OUP user menu

Cholesterol levels in small LDL particles predict the risk of coronary heart disease in the EPIC-Norfolk prospective population study

Benoit J. Arsenault, Isabelle Lemieux, Jean-Pierre Després, Nicholas J. Wareham, Robert Luben, John J.P. Kastelein, Kay-Tee Khaw, S. Matthijs Boekholdt
DOI: http://dx.doi.org/10.1093/eurheartj/ehm390 2770-2777 First published online: 17 October 2007


Aims To evaluate the association of low-density lipoprotein cholesterol (LDL-C) levels in small and large LDL particles with risk of incident coronary heart disease (CHD).

Methods and results We performed a prospective case-control study nested in the EPIC-Norfolk cohort. Cases were apparently healthy men and women aged 45–79 years who developed fatal or non-fatal CHD (n = 1035), and who were matched by age, gender, and enrollment time to 1920 controls who remained free of CHD. Electrophoretic characteristics of LDL particles were measured using 2–16% polyacrylamide gradient gel electrophoresis. Concentrations of LDL-C<255 Å were higher in cases than controls in men (1.34 ± 0.88 vs. 1.15 ± 0.80 mmol/L, P < 0.001) as well as in women (1.12 ± 0.84 vs. 0.94 ± 0.74 mmol/L, P < 0.001). The unadjusted odds ratio (OR) for future CHD in men of the top tertile of LDL-C<255 Å was 1.68 (95% CI, 1.33–2.13; P < 0.001) whereas in women the unadjusted OR was 1.53 (95% CI, 1.13–2.07; P < 0.001). However, after further adjustments for confounding variables, the association between LDL-C<255 Å and CHD was no longer significant in men and in women.

Conclusion Cholesterol concentrations in different LDL subclasses show different relationships with CHD risk in this European cohort.

  • Epidemiology
  • Gender
  • Lipids
  • Risk factors
  • LDL particle size


Among metabolic risk factors of coronary heart disease (CHD), low-density lipoprotein cholesterol (LDL-C) represents the lipoprotein–cholesterol fraction which is the primary target of therapy.1 Over the last decades, numerous epidemiological studies have linked LDL-C to CHD morbidity and mortality and a range of randomized controlled trials have shown the clinical benefits of lowering LDL-C with HMG-CoA reductase inhibitors (statins).25 However, in Western societies, many individuals with LDL-C levels apparently within the normal range have visceral obesity and are characterized by a typical atherogenic dyslipidemia.6 These individuals have low levels of high-density lipoprotein cholesterol (HDL-C), high levels of triglycerides, and apolipoprotein B (apoB), as well as an increased preponderance of small, dense LDL particles, and are at a considerably increased risk of CHD.7 Moreover, at any level of LDL-C, individuals with an elevated proportion of small LDL particles have been reported to be at greater risk for CHD compared with individuals with the same LDL-C levels, but with larger LDL particles.8,9 It is now well recognized that LDL particles are heterogeneous in terms of lipid composition, size, and density and that among viscerally obese, dyslipidemic individuals, physicochemical properties of LDL particles are altered.10,11 Owing to their longer plasma half-life, small and dense LDL particles are more vulnerable to oxidation and glycosylation, and as a consequence, may be more atherogenic.12 Several studies have suggested that an elevated concentration of small and dense LDL particles is predictive of an increased risk of CHD.13,14 The measurement of LDL-C levels within different LDL subclasses could therefore represent a promising method to combine markers of LDL quantity and quality in the assessment of CHD risk.

Most prospective case–control studies evaluating the association of LDL size and LDL particle number with cardiovascular events have reported a significant relationship of CHD with either small LDL particles and/or an elevated concentration of LDL particles in univariate analyses.15 However, adjustment for correlates of LDL particle size such as triglyceride levels, HDL-C levels, or other features of the metabolic syndrome was usually found to attenuate this association.16,17 The results of these studies are limited by the fact that most were performed among men only, investigated a small number of participants, or had a short follow-up.

The primary objective of the present study was therefore to investigate whether the proportion of small and large LDL particles, and especially the cholesterol levels within small and large LDL subfractions, was associated with an increased risk of CHD. We tested whether these features of the small LDL phenotype were predictive of CHD risk in apparently healthy men and women of a European cohort representative of a contemporary Western population.


Study design

We performed a nested case–control analysis of participants of the European Prospective Investigation into Cancer and Nutrition (EPIC)-Norfolk study, a prospective population study of 25 663 men and women aged between 45 and 79 years, residing in Norfolk, UK, who completed a baseline questionnaire survey and attended a clinical visit.18 Participants were recruited from age–sex registers of general practices in Norfolk as part of the nine-country collaborative EPIC study designed to investigate dietary and other determinants of cancer. Additional data were obtained from EPIC-Norfolk to enable the assessment of determinants of other diseases. The study cohort was closely similar to the UK population samples with regards to many characteristics, including anthropometry, blood pressure, and lipids but with a lower proportion of smokers.18

The design and methods of the study have been described in detail elsewhere.18 In short, eligible participants were recruited by mail. At the baseline survey between 1993 and 1997, participants completed a detailed health and lifestyle questionnaire. Blood was taken by venipuncture into plain and citrate tubes. Blood samples were processed for various assays at the Department of Clinical Biochemistry, University of Cambridge, or stored at −80°C. All individuals have been flagged for mortality at the UK Office of National Statistics, with vital status ascertained for the entire cohort. Death certificates for all decedents were coded by trained nosologists according to the International Classification of Diseases (ICD) 9th revision. Death was considered due to CHD when the underlying cause was coded as ICD 410–414. These codes encompass the clinical spectrum of CHD, i.e. unstable angina, stable angina, and myocardial infarction. In addition, participants admitted to hospital were identified by their unique National Health Service number by data linkage with ENCORE (East Norfolk Health Authority database), which identifies all hospital contacts throughout England and Wales for Norfolk residents. Participants were identified as having CHD during follow-up if they had a hospital admission and/or died with CHD as an underlying cause. The Norwich District Health Authority Ethics Committee approved the study, and all participants gave signed informed consent.


For the present nested case–control study, 1035 apparently healthy individuals who ultimately developed fatal or non-fatal CHD during follow-up were identified. Apparently healthy individuals were defined as study participants who did not report a history of heart attack or stroke at the baseline clinic visit. Consequently, we excluded all individuals reporting a history of heart attack or stroke at the baseline clinical visit. Control participants were apparently healthy study participants who remained free of any cardiovascular disease during follow-up. Two controls were matched to each case by sex, age (within 5 years), and date of visit (within 3 months). A total of 1920 controls could be matched to cases. We report results with follow-up till January 2003, with an average of 7.7 ± 1.1 years.

Biochemical analyses

Non-fasting serum levels of total cholesterol, HDL-C, and triglycerides were measured on fresh samples with the RA 1000 (Bayer Diagnostics, Basingstoke, UK), and LDL-C levels were calculated using the Friedewald formula.19 Serum levels of apoB were measured by rate immunonephelometry (Behring Nephelometer BNII, Marburg, Germany) with calibration traceable to the International Federation of Clinical Chemistry primary standards.20 C-reactive protein (CRP) levels were measured with a sandwich-type ELISA in which polyclonal rabbit anti-CRP protein antibodies were used as catching antibodies and biotinylated monoclonal antibodies (CLB anti-CRP-2) as the detecting antibody, as described previously.21 The results were related to a standard consisting of commercially available CRP (Behringwerke AG, Marburg, Germany). The lower detection limit was 0.1 mg/L.

Low-density lipoprotein particle size characterization

In order to measure LDL particle size, non-denaturing 2–16% polyacrylamide gradient gel electrophoresis (PAGGE) was performed on whole plasma kept at −80°C.22,23 Gels were made with acrylamide and bis-acrylamide (37.5:1) purchased from Bio-Rad (Hercules, CA, USA) and were cast in our laboratory. A volume of 3 µL of whole plasma and 3 µL of sample buffer made from sucrose (20%) and bromophenol blue (0.25%) was loaded onto the gels. Electrophoresis was performed in a refrigerated cell (4°C) for a pre-run of 15 min at 125 V. This pre-run preceded electrophoresis of the plasma samples at 70 V for 20 min and at 150 V for 4 h. Gels were stained with Sudan black solution for lipids overnight (7 g/100 mL). Destaining of the gels was performed in a 20% ethanol solution for 24 h and original gel size was restored overnight in a 4% methanol and 0.8% acetic acid solution. Gels were then scanned using an Alpha Imager scanner from Fisher Scientific (Nepean, Ontario, Canada) and were analysed with the Imagemaster 1D Prime v4.10 computer software (Amersham Pharmacia Biotech). LDL size was extrapolated from the relative migration of three plasma standards of known diameters (238.3–274.2 Å). The estimated diameter for the major peak of each sample was identified as the LDL peak particle size. The relative proportion of LDL having a diameter of <255 or >260 Å was ascertained by computing the relative area of the densitometric scan <255 and >260 Å. It has been documented that Sudan black stains mainly non-polar lipids.24 The absorbance profile with Sudan black staining was also assumed to closely reflect the cholesterol distribution among LDL particles of different sizes.25 The absolute concentration of cholesterol among particles of size <255 Å was calculated by multiplying the plasma LDL-C levels by the relative proportion of LDL with a diameter of <255 Å. A similar approach was used to assess the relative and absolute LDL-C levels in particles with a diameter of >260 Å. Samples were analysed in random in order to avoid systemic bias. Researchers and laboratory personnel had no access to identifiable information and could identify samples by number only.

Statistical analyses

Baseline characteristics were compared between cases and controls taking into account the matching between them. A mixed effect model was used for continuous variables and conditional logistic regression was used for categorical variables. Because triglyceride and CRP levels had a skewed distribution, values were log-transformed before being used as continuous variables in statistical analyses. To determine relationships between LDL characteristics, cardiovascular risk factors, and the risk of CHD, we calculated mean risk factor levels for all LDL particle characteristic tertiles. Tertiles were based on the distribution found in control participants. Mean levels of traditional cardiovascular risk factors were calculated either per tertile of the proportion of small or large LDL or per tertile of LDL-C in small or large particles. For sex-specific analyses, sex-specific tertiles were used, and for pooled analyses, we used tertiles based on the combined sexes. Conditional logistic regression analysis was used to calculate odds ratio (OR) and corresponding 95% CI as an estimate of the relative risk of CHD. ORs were calculated taking into account the matching for sex and age and were adjusted for smoking (never, past, and current) and further adjusted for the following cardiovascular risk factors: diabetes (yes/no), body mass index (BMI), systolic blood pressure (SBP), and LDL-C. ORs were also calculated after additional adjustment for HDL-C and log-transformed triglycerides. To assess whether LDL particle size and subclasses had predictive value on top of the Framingham risk score (FRS), we calculated ORs for future CHD per LDL subclass tertile, simultaneously adjusting for the FRS in the model. The FRS was calculated using a previously reported algorithm, which takes into account age, sex, total cholesterol, HDL-C, SBP, diastolic blood pressure, smoking, and the presence of diabetes.26 The calculated FRS was subsequently categorized into three groups: low (<10%), intermediate (10–20%) and high (>20%) risk. Statistical analyses were performed using SPSS software (Version 12.0.1, Chicago, IL, USA). A P-value of <0.05 was considered statistically significant.


Characteristics of participants are presented in Table 1. Matching ensured that age was not different between cases and controls. A total of 885 cases could be matched to two controls, whereas 150 cases could be matched to one control only. In both men and women, cases of CHD had higher adiposity indices such as BMI and waist circumference than controls (P < 0.001 for each). Cases also had higher SBP and diastolic blood pressure and were more likely to smoke and to have diabetes (P ≤ 0.001). Men and women who developed CHD during follow-up also had a plasma lipoprotein–lipid profile comprising lower HDL-C levels and elevated triglyceride and LDL-C concentrations (P ≤ 0.005 for each). Participants who developed CHD during follow-up were characterized by a smaller LDL peak particle size compared with those who remained CHD free (P ≤ 0.001) and the proportion of small LDL particles was higher in cases than controls in men and women (P ≤ 0.001), whereas the proportion of large LDL particles was lower among cases of CHD than controls (P ≤ 0.001). Figure 1 shows that participants with CHD were also characterized by increased LDL-C<255 Å levels compared with controls (P < 0.001) while LDL-C>260 Å levels were not statistically different between the two groups.

Figure 1

LDL-C levels in small (top) and large (bottom) LDL particles of cases and controls in men and women. NS, not significant.

View this table:
Table 1

Baseline characteristics of 2955 men and women included in EPIC-Norfolk, 1993–2003

Controls (n = 1209)Cases (n = 660)Controls (n = 711)Cases (n = 375)
Age, years65 ± 865 ± 867 ± 767 ± 7
Smoking, %
BMI, kg/m226.3 ± 3.127.2 ± 3.5*26.1 ± 3.927.3 ± 4.6*
Waist circumference, cm95.6 ± 9.398.3 ± 9.9*83.2 ± 10.186.5 ± 11.3*
History of diabetes, % (n)2.1 (25)7.0 (46)*1.0 (7)5.1 (19)*
Systolic blood pressure, mmHg139 ± 18145 ± 19*138 ± 19142 ± 19*
Diastolic blood pressure, mmHg84 ± 1187 ± 12*82 ± 1184 ± 11*
Total cholesterol, mmol/L6.0 ± 1.16.3 ± 1.1*6.6 ± 1.26.8 ± 1.2**
LDL-C, mmol/L4.0 ± 1.04.1 ± 1.0*4.3 ± 1.14.5 ± 1.1**
HDL-C, mmol/L1.3 ± 0.31.2 ± 0.3*1.6 ± 0.41.4 ± 0.4*
Triglycerides, mmol/L1.7 (1.2–2.3)1.9 (1.4–2.7)*1.5 (1.1–2.1)1.8 (1.3–2.4)*
ApoB, mg/dL126 ± 29136 ± 30*131 ± 32139 ± 34*
C-reactive protein, mg/L1.4 (0.7–2.9)2.1 (1.0–4.5)*1.6 (0.7–3.4)2.6 (1.1–5.9)*
LDL peak particle size, Å260 ± 5259 ± 5*262 ± 5261 ± 5*
% LDL<255 Å28.7 ± 17.932.5 ± 19.3*21.7 ± 15.125.1 ± 16.8*
% LDL>260 Å44.1 ± 18.741.2 ± 19.2*53.1 ± 17.948.9 ± 19.2*
  • Data are presented as mean (±SD), % (n), or median (interquartile range).

  • BMI, body mass index; LDL-C, low-density lipoprotein cholesterol; HDL-C, high-density lipoprotein cholesterol; apoB, apolipoprotein B; % LDL<255 Å, proportion of LDL with a diameter of <255 Å; % LDL>260 Å, proportion of LDL with a diameter of >260 Å.

  • *P ≤ 0.001 and **P ≤ 0.005 (significantly different from the corresponding subgroup).

In order to further explore the relationship between LDL-C<255 Å and characteristics of participants, men and women were classified according to this variable. Table 2 shows characteristics of participants according to tertiles of LDL-C<255 Å. In both men and women, participants with the highest LDL-C<255 Å concentrations were characterized by a more adverse metabolic risk profile. Participants in the top tertile had elevated concentrations of triglycerides, CRP (in men only) and apoB while being characterized by the lowest HDL-C levels (P ≤ 0.001).

View this table:
Table 2

Characteristics of men and women according to tertiles of LDL-C in small LDL particles (LDL-C<255Å, mmol/L) in EPIC-Norfolk, 1993–2003

MenTertiles of LDL-C<255 Å (mmol/L)P-Value
<0.67 (0.39 ± 0.18)0.67–1.34 (0.98 ± 0.19)>1.34 (2.10 ± 0.64)
Cases–controls (%)174/402 (43.3)187/403 (46.4)299/404 (74.0)
Age, years65 ± 865 ± 864 ± 80.05
Smoking, %
BMI, kg/m226.1 ± 3.226.4 ± 3.427.1 ± 3.2<0.001
Waist circumference, cm95 ± 1096 ± 1098 ± 9<0.001
History of diabetes, %(n)4.2 (24)4.1 (24)3.3 (23)NS
Systolic blood pressure, mmHg139 ± 18141 ± 18143 ± 18<0.001
Diastolic blood pressure, mmHg84 ± 1185 ± 1286 ± 11<0.001
Total cholesterol, mmol/L5.8 ± 1.05.9 ± 1.06.6 ± 1.0<0.001
LDL-C, mmol/L3.7 ± 1.03.9 ± 0.94.4 ± 0.9<0.001
HDL-C, mmol/L1.3 ± 0.41.3 ± 0.31.1 ± 0.3<0.001
Triglycerides, mmol/L1.6 ± 0.71.7 ± 0.72.3 ± 0.9<0.001*
ApoB, mg/dL117 ± 28123 ± 24146 ± 27<0.001
C-reactive protein, mg/L3.0 ± 5.13.5 ± 5.43.9 ± 6.6<0.001*
LDL peak particle size, Å263.4 ± 3.2260.0 ± 3.0255.7 ± 3.7<0.001
% LDL<255 Å11.3 ± 6.826.5 ± 8.148.4 ± 13.4<0.001
% LDL>260 Å63.4 ± 12.843.7 ± 9.725.9 ± 10.4<0.001
LDL-C>260 Å, mmol/L2.39 ± 0.861.73 ± 0.671.17 ± 0.58<0.001
Women<0.52 (0.29 ± 0.14)0.52–1.07 (0.78 ± 0.16)>1.07 (1.79 ± 0.68)P-Value
Cases–controls, %112/237 (47.3)89/236 (37.7)174/238 (73.1)
Age, years66 ± 766 ± 767 ± 7NS
Smoking, %
BMI, kg/m226.1 ± 4.026.3 ± 4.227.1 ± 4.30.003
Waist circumference, cm82 ± 1084 ± 1186 ± 11<0.001
History of diabetes, %(n)1.1 (4)2.5 (8)3.4 (14)NS
Systolic blood pressure, mmHg137 ± 17139 ± 19142 ± 200.001
Diastolic blood pressure, mmHg81 ± 1083 ± 1284 ± 120.002
Total cholesterol, mmol/L6.4 ± 1.16.4 ± 1.17.1 ± 1.2<0.001
LDL-C, mmol/L4.1 ± 1.04.1 ± 1.04.8 ± 1.1<0.001
HDL-C, mmol/L1.7 ± 0.41.6 ± 0.41.4 ± 0.4<0.001
Triglycerides, mmol/L1.5 ± 0.61.6 ± 0.72.1 ± 0.9<0.001*
ApoB, mg/dL124 ± 27125 ± 29149 ± 34<0.001
C-reactive protein, mg/L3.8 ± 6.14.3 ± 7.14.0 ± 6.1NS*
LDL peak particle size, Å265.0 ± 3.2262.0 ± 3.0257.6 ± 3.9<0.001
% LDL<255 Å7.4 ± 4.119.9 ± 6.638.3 ± 12.6<0.001
% LDL>260 Å70.9 ± 10.752.6 ± 9.334.6 ± 11.3<0.001
LDL-C>260 Å, mmol/L2.92 ± 0.952.19 ± 0.791.68 ± 0.75<0.001
  • Data are presented as mean (±SD) or %(n).

  • *On log-transformed values.

Figure 2 represents ORs for future CHD according to tertiles of LDL-C<255 Å (Figure 2A) and LDL-C>260 Å (Figure 2B). The OR for CHD was significantly increased by 1.68-fold (95% CI, 1.33–2.13, P < 0.001) among men in the top tertile of LDL-C<255 Å compared with men of the first tertile. This relationship remained significant after adjustment for classical risk factors such as age, smoking, diabetes, BMI, SBP, and LDL-C (OR, 1.43; 95% CI, 1.10–1.85, P = 0.005). However, statistical significance was lost after further adjustment for triglyceride and HDL-C levels (OR, 1.20; 95% CI, 0.91–1.59, NS). Unadjusted OR for CHD among women in the top LDL-C<255 Å tertile compared with women in the first tertile was 1.53 (95% CI, 1.13–2.07; P = 0.003) and 1.54 (95% CI, 1.14–2.10, P = 0.003) after adjustment for age and smoking. Further adjustment for diabetes, BMI, SBP, and LDL-C attenuated the relationship between LDL-C<255 Å and CHD (OR, 1.29; 95% CI, 0.92–1.79, NS). Finally, there was no significant association between tertiles of LDL-C>260 Å and CHD in both men and women.

Figure 2

Univariate and multivariate odds ratio (OR) of CHD associated with LDL-C levels in small (A) and large (B) LDL particles of men and women. Black circles represent unadjusted OR. White squares represent OR after adjustment for age and smoking. Black triangles represent OR after adjustment for classical risk factors (age, smoking, diabetes, BMI, systolic blood pressure, and LDL-C), and black squares represent OR after adjustment for HDL-C and log-transformed triglycerides.

Kaplan–Meier survival curves according to tertiles of proportion of small and large LDL particles as well as for tertiles of LDL-C<255 Å and LDL-C>260 Å were assessed in the entire cohort. Figure 3 shows that participants characterized by an increased proportion of small LDL particles or participants with elevated LDL-C<255 Å concentrations had poorer survival probabilities than participants with a low proportion of small LDL particles or a low concentration of LDL-C<255 Å (P < 0.001). On the other hand, participants with an increased proportion of large LDL particles had better survival probabilities (P < 0.001). However, the reverse association between LDL-C>260 Å and CHD did not reach statistical significance (P = 0.07).

Figure 3

Kaplan–Meier survival curves according to tertiles of the proportion of small (A) and large LDL particles (B), and according to LDL-C levels in small (C) and large (D) particles in men and women over the follow-up. P-values were not adjusted.

Finally, our indices of electrophoretic properties of LDL were classified into tertiles or examined as continuous variables to explore whether various characteristics of LDL particles had a predictive value on top of the FRS in predicting the risk of CHD. Table 3 shows that when characteristics of LDL particles are studied as tertiles or used as continuous variables, the relationship between the percentage of small LDL particles or the LDL-C levels in small particles (<255 Å) and CHD was independent of the FRS (P = 0.02).

View this table:
Table 3

Odds ratios for future CHD in 1035 cases and 1920 controls after adjustment for Framingham risk score by indices of LDL characteristics as tertiles and as continuous variables in EPIC-Norfolk, 1993–2003

% LDL<255 Å
 Unadjusted1.001.08 (0.88–1.31)1.54 (1.27–1.86)<0.0011.12 (1.07–1.17)a<0.001
 Adjusted for FRS1.001.01 (0.83–1.25)1.27 (1.04–1.54)0.021.06 (1.02–1.11)a0.01
% LDL>260 Å
 Unadjusted1.000.82 (0.68–0.98)0.66 (0.55–0.80)<0.0010.91 (0.87–0.95)a<0.001
 Adjusted for FRS1.000.94 (0.77–1.13)0.82 (0.67–1.01)NS0.96 (0.92–1.00)aNS
LDL-C<255 Å, mmol/L
 Unadjusted1.001.17 (0.95–1.43)1.72 (1.42–2.09)<0.0011.33 (1.21–1.46)b<0.001
 Adjusted for FRS1.001.11 (0.90–1.36)1.33 (1.09–1.63)0.0051.13 (1.03–1.25)b0.02
LDL-C>260 Å, mmol/L
 Unadjusted1.000.82 (0.68–0.98)0.85 (0.70–1.02)NS0.95 (0.87–1.03)bNS
 Adjusted for FRS1.000.86 (0.71–1.04)0.90 (0.74–1.09)NS0.97 (0.89–1.06)bNS
  • FRS, Framingham risk score.

  • aOdds ratios for tertiles of proportions of small and large particles are per 10% points.

  • bOdds ratios for LDL-C levels in small and large LDL particles as continuous variables are per 1 mmol/L.


To the best of our knowledge, the present study is the first prospective population-based study to compare cholesterol levels within different LDL subclasses to the risk of developing CHD in both men and women over a relatively long follow-up period. It is now clear that the small LDL phenotype is a feature of the atherogenic dyslipidemia mainly associated with hypertriglyceridemia, low HDL-C levels, and visceral obesity.27 In the EPIC-Norfolk prospective population study, we found that both men and women who developed CHD during follow-up had a smaller LDL peak particle size than matched controls. These participants were also characterized by increased cholesterol levels associated with their small LDL particles. This relationship between LDL-C<255 Å of men and the risk of CHD remained significant even after adjustments for traditional risk factors such as age, smoking, diabetes, BMI, SBP, and LDL-C. In women, this relationship was no longer statistically significant after such adjustment. In both sexes, however, the relationship between LDL-C<255 Å and CHD was no longer significant after further adjustment for triglyceride and HDL-C levels. Survival curves also support the notion that men and women with elevated LDL-C<255 Å concentrations have reduced survival probabilities, whereas participants with high LDL-C>260 Å levels did not appear to be at increased risk for CHD. In fact, these participants rather appeared to be somewhat protected against CHD.

Cholesterol levels of low-density lipoprotein subclasses and risk of coronary heart disease

A negative association has been reported between the size of LDL particles and the progression of angiography-documented atherosclerosis28 and the incidence of clinical coronary artery disease events.29 In the Québec Cardiovascular Study, the association between the small LDL phenotype and the risk of ischaemic heart disease (IHD) was independent of the concomitant variations in lipid and non-lipid variables such as HDL-C and triglyceride levels.14 Additional analyses by St-Pierre et al.23 suggested that further characterization of LDL particles’ physicochemical properties using PAGGE to determine LDL-C<255 Å could better predict the risk of IHD in a 5-year follow-up. In the same cohort, similar analyses performed over a 13-year follow-up showed that such relationship between LDL-C<255 Å levels and the risk of IHD was particularly evident during the first half of the follow-up, suggesting that the small, dense LDL phenotype may be helpful to predict early events and less useful to predict long-term CHD risk.30

Results of the present study also support the notion that elevated LDL-C<255 Å concentrations are associated with an increased risk of CHD, but in contrast with the Québec Cardiovascular Study, we could not find an association between features of the small LDL phenotype and CHD that was independent of triglyceride and HDL-C levels. This expected finding supports the notion that small and dense LDL, hypertriglyceridemia, and low levels of HDL-C are part of an atherogenic lipoprotein phenotype that might be features of a broader ‘dysmetabolic’ profile linked to abdominal obesity, insulin resistance, and the metabolic syndrome. Our study has the advantage of presenting results for both men and women while the Québec Cardiovascular Study and other prospective studies on this topic were mostly performed in men. Our finding is in accordance with most of the cross-sectional and prospective, case–control studies on this topic such as those performed by Stampfer et al.16 and by Austin et al.17

Pathophysiology of small and dense low-density lipoproteins

From a pathophysiological point of view, visceral adipose tissue is an important source of non-esterified fatty acids that use the portal vein to reach the liver, which, in turn, contribute to increase the secretion of triglyceride-rich very low-density lipoproteins.31 Moreover, another consequence of hypertriglyceridemia is the cholesteryl ester transfer protein (CETP)-mediated transfer of triglycerides from triglyceride-rich lipoproteins to LDL and HDL particles. In exchange, triglyceride-rich lipoproteins accept cholesteryl esters from LDL and HDL particles. This process contributes to triglyceride enrichment and cholesteryl ester depletion of apoB-containing lipoproteins, which upon triglyceride hydrolysis by various lipases makes them smaller, denser, and presumably, more atherogenic.32 CETP activity is the principal factor to explain why triglyceridemia and the small and dense LDL phenotype are so closely associated.33 Moreover, because of their poor affinity towards the LDL receptor, small and dense LDL particles have a longer residence time than larger, more buoyant particles and, as a consequence, are more likely to undergo certain chemical modifications in vivo.9 It is interesting to point out that dyslipidemic and abdominally obese subjects often have elevated reactive oxygen species (ROS) and blood glucose levels.34,35 There is considerable evidence showing that small and dense LDL particles are particularly vulnerable to oxidation and glycosylation.36,37 Accumulating data also suggest that apoB, phospholipids, cholesteryl esters, and free cholesterol are components of LDL particles that are good targets for ROS and other non-radical oxidative species.38 On the other hand, chronic exposition to hyperglycemia leads to the formation of advanced glycation end products which, like ROS, have the potential to modify the native form of LDL particles and to accentuate their atherogenic potential.35,39 However, oxidized and glycosylated LDL particles were not measured in the present study.


Certain aspects of this study merit further consideration. First, it is important to point out that electrophoretic characteristics of LDL particles were determined in non-fasting samples. The fact that these samples were collected at different periods of the day could influence triglyceride and LDL-C levels. Since triglyceride levels are the best correlate of LDL particle physicochemical characteristics, this could have had an impact on LDL properties, although we40 and other groups41 have shown that post-prandial lipemia does not have a strong effect on LDL particle size. Secondly, PAGGE was performed using samples that had been kept at −80°C for several years; therefore, we cannot exclude some degree of protein and/or membrane degradation. Moreover, small LDL particles have a reduced proportion of esterified cholesterol as opposed to larger particles. With our technique, we have to assume each and every LDL particle, independent of its size, carries the same proportion of cholesterol to evaluate the cholesterol levels within small and large LDL particles. These limitations would introduce an increased random measurement error, which is likely to lead to an underestimation of any relationship. Most of participants were somewhat advanced in age at the beginning of the follow-up, which could also introduce a survival bias. For instance, Barzilai et al.42 have shown that an increased LDL particle size was associated with longevity among Ashkenazi Jewish probands with exceptional longevity. However, truncation of the distribution would reduce power to detect the associations.


Results of the present study suggest that an increased concentration of small LDL particles (increased LDL-C<255 Å levels) is predictive of an increased risk of CHD in apparently healthy men and women. Individuals with the highest cholesterol levels in their small LDL particles were characterized by an atherogenic metabolic risk profile and were at greater risk of CHD than individuals characterized by lower LDL-C<255 Å concentrations. However, this relationship was not totally independent from concomitant variations in lipid levels. Relationships were similar among men and women. This prospective study raises the question of the importance of measuring LDL particle size and physicochemical properties in clinical practice. In our view, the presence of abdominal fat (as measured by computed tomography scan or indirectly by assessment of waist circumference) along with elevated triglyceride levels and decreased HDL-C concentrations should provide enough information in assessing the risk of CHD related to small LDL in clinical practice. Further studies are clearly needed in order to better evaluate the contribution of LDL characteristics to the development of CHD, especially in patients with the metabolic syndrome. Finally, the lack of an independent relationship between LDL physicochemical properties and CHD does not exclude a pathophysiological role for these particles in the atherosclerotic process.


EPIC-Norfolk is supported by program grants from the Medical Research Council UK and Cancer Research UK and with additional support from the European Union, Stroke Association, British Heart Foundation, Department of Health, Food Standards Agency and the Wellcome Trust.


We would like to thank Ms. Patricia Blackburn for her critical review of this manuscript. We also thank participants, general practitioners, and staff in EPIC-Norfolk. None of the study sponsors have had any role in study design, collection analysis and interpretation of data, and writing of the report or decision to submit the paper for publication. Jean-Pierre Després is the Scientific Director of the International Chair on Cardiometabolic Risk which is supported by an unrestricted grant awarded to Université Laval by Sanofi Aventis.

Conflict of interest: none declared.


View Abstract