Clinica Chimica Acta 413 (2012) 251–257

Contents lists available at SciVerse ScienceDirect

Clinica Chimica Acta

j ourna l homepage: www.e lsev ie r .com/ locate /c l inch im

Comparability of two different polyacrylamide gel electrophoresis methods for theclassification of LDL pattern type

Celia Bañuls a,b, Lorena Bellod a,b, Ana Jover a,b, Maria Luisa Martínez-Triguero c, Víctor Manuel Víctor a,b,d,e,Milagros Rocha a,b,e, Antonio Hernández-Mijares a,b,f,⁎a Service of Endocrinology, University Hospital Dr. Peset, Valencia, Spainb Dr. Peset Hospital Research Foundation, Valencia, Spainc Unit of Clinical Analysis, University Hospital La Fe, Valencia, Spaind Department of Physiology, Faculty of Medicine, University of Valencia, Spaine CIBER CB06/04/0071 research group, CIBER Hepatic and Digestive Diseases, University of Valencia, Spainf Department of Medicine, Faculty of Medicine, University of Valencia, Spain

⁎ Corresponding author at: Service of Endocrinology, UGaspar Aguilar 90, 46017 Valencia, Spain. Tel.: +34 96162

E-mail address: hernandez_antmij@gva.es (A. Herná

0009-8981/$ – see front matter © 2011 Elsevier B.V. Alldoi:10.1016/j.cca.2011.09.047

a b s t r a c t

a r t i c l e i n f o

Article history:

Received 23 June 2011Received in revised form 29 September 2011Accepted 30 September 2011Available online 6 October 2011

Keywords:Small, dense LDLCardiovascular riskPolyacrylamide gradient gel electrophoresisPolyacrylamide gradient tube electrophoresisTriglycerides

Background: The measurement of small dense low-density lipoprotein (sdLDL) particles is relevant whenassessing cardiovascular risk. However, there is as yet no referenced method for the determination of LDLsubfractions or a standardized comparison of the methods currently available.Therefore, the aim of this study was to compare the pattern of LDL particles measured by polyacrylamidetube gel electrophoresis (PTGE) and polyacrylamide gradient gel electrophoresis (PGGE) and to correlatethe results with triglyceride concentration.Materials and methods: Serum samples were collected from 177 patients. Lipid profile and LDL particle sizewere assessed using PTGE and PGGE.Results: Pearson correlation and kappa index revealed a very good agreement between the methods. Therewas 81.3% concordance for classification of sdLDL particles and 97.2% concordance for classification of largeLDL when PTGE and PGGE were compared. LDL size correlated with triglyceride in subjects with triglyceride

levels N116 mg/dl, pointing to a high CAD risk, as reflected by their higher prevalence of pattern B.Conclusions: PTGE correlates favourably and is in very good agreement with PGGE. The determination of LDL par-ticle sizemay be an appropriate analytical procedure to estimate CAD risk in patients with high triglyceride levels.

© 2011 Elsevier B.V. All rights reserved.

1. Introduction

The risk of atherosclerosis has been related with abnormalities inplasma lipoprotein concentrations. Specifically, a connection has beendemonstrated between high plasma levels of low-density lipoprotein(LDL) cholesterol and an increased risk of coronary heart disease (CHD)[1]. Thus, it is logical that LDL cholesterol is the primary therapeutic tar-get for coronary artery disease (CAD) prevention. Nevertheless, an in-creasing amount of research over the past decade has been devoted tothe heterogeneity of LDL particles and the atherogenicity of lipids and li-poproteins other than the LDL type. It iswell known that LDL particles area heterogeneous population that varies with respect to lipoprotein com-position, size and metabolism. LDL particle size has been the focus of at-tention because of their atherogenic potential. Typically, individualswithlarge LDL particles are classified as pattern A, whereas individuals withsmall and dense LDL (sdLDL) particles are classified as pattern B and

niversity Hospital Dr. Peset, Av.2492; Fax: +34 961622492.ndez-Mijares).

rights reserved.

are at a greater risk of developing CAD [2,3]. Therefore, the predominanceof sdLDL particles is considered an emerging risk factor for CAD by theNational Cholesterol Education Program Adult Treatment Panel III [4].These particles are more easily concentrated in arterial walls, are moreprone to oxidation, and have a reduced affinity for LDL receptors whencompared with larger LDL particles [5].

When estimating cardiovascular disease (CVD) risk measurement ofLDL subfractionsmay provide additional predictive power to LDL choles-terol concentration measurement, alone or in combination with assess-ment of other risk factors. For instance, individuals with smaller LDLparticle size will be at higher risk of CAD than those with a similarnormal/borderline elevated LDL cholesterol concentration but largerLDL particle size. Therefore, in these cases, the risk of atherosclerosis isbetter indicated by LDL particle size [6]. Despite the established role ofsdLDL particles in the development of atherosclerosis, there is no refer-ence method for determining LDL subfractions, since currently availableanalytical approached require expensive instrumentation and are time-and labour-intensive, and thus are not easily applicable in routine clinicalpractice [7]. Currently, methods for separating LDL subfractions includedensity gradient ultracentrifugation [8], nuclear magnetic resonance

252 C. Bañuls et al. / Clinica Chimica Acta 413 (2012) 251–257

[9], and non-denaturing polyacrylamide gradient gel electrophoresis(PGGE) [10]. Unfortunately, there is no standardization among the differ-ent methods employed, and comparability in terms of results is not ex-trapolated. In fact, the results of a systematic review have raisedcontroversy, finding a wide range of variability of concordance (from 7to 94%) in the classification of LDL particle patterns [11]. An alternativeand less demanding approach to the study of LDL subclasses is amodifiedPGGE technique called polyacrylamide tube gel electrophoresis (PTGE),which has been commercialised by the Lipoprint® System (Quantime-trix, Redondo Beach, CA, USA) [12]. It measures particle size and allowssdLDL fractions to be quantified, and is relatively simple to operate andfast when compared with the PGGE. It is approved by the Food andDrug Administration (FDA) and uses a standardized method to separateLDL particles into 7 subfractions on the basis of size. As the gels are pre-pared by the manufacturer, it is technically simpler, less resource-intensive, and more conducive to routine laboratory testing thantraditional gel electrophoresis [13].

Only a few studies have compared PTGE and PGGE [12–14], andconcordance rates vary according to LDL patterns. Therefore, theaim of this study was to compare the pattern of LDL particles mea-sured by PTGE on the one hand and PGGE on the other and to associ-ate the results of the former method with lipid parameters of CVD riskthat are known to be affected by LDL particle size.

2. Materials and methods

2.1. Subjects

Patients were recruited in the Service of Endocrinology and Nutri-tion of University Hospital Dr Peset (Valencia, Spain). Blood samplesfrom 177 patients were collected after 12 h overnight fasting. Patientswere excluded if they presented a neoplastic, renal or liver disease orwere suffering from a recent trauma or major surgery. Pregnant andlactating women were also excluded.

In order to separate serum from blood cells, venous blood sampleswere centrifuged at 2000 g for 15 min at 4 °C. Freshly separatedserumwas employed to determine lipid profile, and the remaining al-iquots of serum were stored at −80 °C until determination of LDLsubfractions by the two aforementioned methods in our Lipid Re-search Unit. The study was conducted according to the guidelineslaid down in the Declaration of Helsinki, and all procedures involvinghuman subjects were approved by our hospital's Ethics Committee.Written informed consent was obtained from all patients.

2.2. Lipid parameters

Total cholesterol and triglycerides weremeasured bymeans of enzy-matic assays [15,16], and high-density lipoprotein (HDL) cholesterolconcentrations were recorded with a Beckman LX-20 autoanalyzer(Beckman Coulter, La Brea, CA, USA) using a direct method [17]. Theintraserial variation coefficient was b3.5% for all determinations. LDLcholesterol concentration was calculated using the method of Friede-wald [18]. Atherogenic index of plasma (AIP) was obtained by calculat-ing the logarithm of the ratio of plasma concentration of triglyceridesto HDL cholesterol. Apolipoprotein B (Apo B) was determined by immu-nonephelometry (Dade Behring BNII, Marburg, Germany) with an intra-assay variation coefficient of b5.5%.

2.3. Electrophoretic methods

2.3.1. Polyacrylamide gradient gel electrophoresisLDL particle sizewas determined by polyacrylamide gradient gel elec-

trophoresis (2–16%), which was freshly cast in the laboratory accordingto the method described by Nichols et al., with slight modifications [19].Serum samples (10 μl), stained with a solution of ethylene-glycol and0.1% (w/v) of Sudan black (Sigma-Aldrich Corporation St. Louis, MO

63178, USA), were applied to the gel in a final concentration of 10% su-crose. Electrophoresis was performed in a refrigerated cell (4–8 °C) fora prerun of 60 min at 120 V, followed by 30 min at 20 V, 30 min at70 V, and 15 h at 90 V. Two standard serums containing 3 LDL fractionswith diameters of 28.35 nm, 26.15 nm and 24.44 nm were used as con-trols. The gels were scanned, and migration distances (from the top ofthe gel to the most prominent band) were measured. The predominantLDL particle diameter of each sample was calculated from a calibrationline. LDL particle subclasses were classified as predominantly small LDLor pattern B (diameter b25.5 nm) and non-small LDL (pattern A, diame-ter ≥25.5 nm) [2]. Both intra- and inter-gel variation coefficients werebelow 1%. Representative appearance of LDL bands is shown in Fig. 1A.

2.3.2. Polyacrylamide tube gel electrophoresisLDL subfractions were separated using the Quantimetrix Lipoprint®

system (Quantimetrix Corporation, Redondo Beach, CA, USA) [12],according to the procedure set out by the manufacturer. In brief, 25 μlof sample were mixed by inversion with 200 μl of Lipoprint loadinggel. This mix was placed on the upper part of precast high resolution3% polyacrylamide gel tubes. After 30 min of photopolymerization atroom temperature in front of a fluorescent light source, samples wereelectrophoresed for 60 min at a constant current of 3 mA for each geltube until the HDL subfraction hadmigrated a distance of approximate-ly 1 cm from the bottom of the tube. Electrophoresis was followed byplacing the tubes in the dark for 30 min before performing densitome-try at 610 nm.

According to electrophoretic mobility based on retardation factor(Rf), very-low density lipoprotein (VLDL) remains in the origin(Rf=0.0), whereas HDL migrates to the front (Rf=1.0). In between,up to 7 LDL bands can bedetected. The LDL1 and LDL2bands correspondwith large, buoyant LDL particles, whereas bands LDL3 to LDL7 corre-spond with small dense LDL particles. The Rf of LDL subfraction isequal to the distance between VLDL and LDL subfraction bands dividedby the distance between VLDL and HDL bands. According to this meth-od, LDL subclasses designated as small have an Rf of more than 0.40,those designated as intermediate have an Rf of 0.38 to 0.40, and thosedesignated as large have an Rf of less than 0.38. Therefore, accordingto the LDL electrophoretic profile, 2 patterns can be defined: patternA, with normal total cholesterol mass of the sdLDL subfractions; andpattern B, in which total cholesterol mass of the sdLDL subfractions isintermediate-low. This system does not measure LDL particle size, butgiven that older systems do, the Lipoprint System provides an estimateby means of the algorithm developed by Kazumi [20]. The Liposure®(Quantimetrix Corporation) was used as quality control.

Raw data obtained from the densitometer were imported and ana-lysedusing a computerizedmethoddeveloped for theQuantimetrix Lipo-print system (Lipoware software). Subfractions were identified andquantified using NIH image program version 1.62 (Bethesda, MD, USA)for research use.

The program partitions bands into discrete segments, and the rel-ative area under the curve is calculated for each lipoprotein band. Italso calculates the cholesterol concentration for each lipoprotein frac-tion according to a total cholesterol value obtained for each sampleusing the aforementioned method.

The average particle size reported by the Lipoprint profile is theweighted average (calculated from the area under the curve for eachsubfraction) of the particle size of all LDL peaks present in the sample.Based on this, the size cut-off works out more than 26.8 nm for patternA (normal LDL size) and to be equal to or less than this threshold forpattern B. Typical migration pattern of large buoyant LDL and smalldense LDL are represented in Fig. 1B and C, respectively.

2.4. Statistical analysis

Statistical analyses were performed using SPSS 15.0 software (SPSSStatistics Inc., Chicago, IL, USA). Continuous variables were expressed as

A

B

C

LDL subfractions

IDLVLDL

HDL

LDL subfractions

IDLVLDL

HDL

Fig. 1. Representative image of LDL bands in PGGE (A) and typical migration pattern of large buoyant LDL (B) and small dense LDL in PTGE (C), with the corresponding scanned andcomputerized lipoprotein profile. A. Lanes 1, 2, 4, 7 and 10 show samples with pattern A, while lanes 3, 8 and 9 show samples with pattern B. Lanes 5 and 6 are two standard serumscontaining 3 LDL fractions with diameters of 28.35 nm, 26.15 nm (C1) and 24.44 nm (C2).

253C. Bañuls et al. / Clinica Chimica Acta 413 (2012) 251–257

mean and standard deviation or asmedian and 25th and 75th percentilesfor parametric and non-parametric data, respectively. Qualitative datawere expressed as percentages.

Comparison of the groups was performed using a one-way ANOVAand Chi-squared test for qualitative variables. Correlation between vari-ables was determined using Spearman's correlation coefficient. Concor-dance between the differentmethods used tomeasure LDL subfractionswas assessed using the kappa index (K). Values of 0.21–0.40, 0.41–0.60,0.61–0.80 and 0.81–1.0 showed fair, moderate, good and very good

concordance, respectively [21]. The individual differences between thetwo methods were calculated and plotted, as described by Bland andAltman [22]. This method assesses the presence of bias when the differ-ence between two methods is not equal across increasing mean levelsof the parameter under study. To further clarify the differences betweenLDL size revealed by PTGE and PGGE, wemodelled the relation betweenthe difference and the mean of both measurements using linear regres-sion. All the tests used a confidence interval of 95%, and differenceswere considered significant when pb0.05.

254 C. Bañuls et al. / Clinica Chimica Acta 413 (2012) 251–257

3. Results

The characteristics of the study population are shown in Table 1. Atotal of 177 subjects (27.1% males, 72.9% females) were selected inorder to provide a wide range of triglyceride and lipid levels, ensuring adistribution of LDL subfractions. Comparison of PTGE and PGGE with re-spect to LDL particle size and classification of different LDL patterns isshown in Fig. 2 and Table 2. In relation to LDL particle size, bothmethodswere positively correlated, providing a correlation coefficient of 0.568,which proved to be highly significant (pb0.001) (Fig. 2A). To detect thebetween-method bias for measuring LDL size, the absolute differencewas plotted against the mean for each pair of measurements (Fig. 2B).The mean difference between PTGE and PGGE was 1.2 nm (with PGGEbeing smaller). The 95% limits of agreement at this point were 0.06 and2.34, indicating that the differences between the two methods can beexpected to fall within this range. The Bland-Altman plot indicates thatthe absolute difference between the twomethods increaseswith increas-ing average LDL size. The linear regression line shown in the figure devi-ated significantly from a horizontal line (pb0.001). In addition to this,classification of LDL pattern was based on electrophoretic mobility forboth methods. For PTGE, pattern B was defined when diameter was ≤26.8 nm and pattern A was confirmed when diameter was N26.8 nm.For PGGE, pattern B and pattern A was defined when diameter was b

25.5 nm and ≥25.5 nm, respectively. PTGE showed an agreement of81.3% concordance for classification of pattern B when compared withPGGE. In the case of pattern A, PTGE had an agreement of 97.2% concor-dance when compared with PGGE. Furthermore, a very good agreementwas detected between PTGE and PGGE after evaluating the data using thekappa statistic (weight K=0.81; very good concordance between 0.81and 1.00 [21]) (Table 2).

Triglyceride and ApoB concentrations and AIP correlated with LDLparticle size when measured by PTGE and PGGE methods (Table 3).Good correlation was confirmed for triglycerides (r=−0.705,pb0.001 and r=−0.484, pb0.001 for PTGE and PGGE, respectively),Apo B (r=−0.532, pb0.001 and r=−0.244, p=0.001 for PTGE andPGGE, respectively) and AIP (r=−0.607, pb0.001 and r=−0.435,pb0.001 for PTGE and PGGE, respectively), although the highest corre-lation coefficients for all parameters were associated with PTGE. For adeeper analysis of correlations, we stratified our population using tri-glyceride tertiles. The results showed that, for triglycerides and AIP, cor-relationwas only significant in subjects of the third tertile; that is, those

Table 1Baseline characteristics and lipid profile of study population.

Mean±SD Range

Sample size (n) 177 —————

Male/Female (n) 48/129 —————

Normolipidemia (%) 51Cases of hyperlipidemia (%)

TCN200 mg/dl 47 —————

TriglyceridesN200 mg/dl 11 —————

TC and triglyceridesN200 mg/dl 8.5Lipid-lowering therapy (%) 10 —————

Cases of metabolic syndrome (%) 26Cases of type 2 diabetes (%) 21 —————

Post-menopausal women (%) 39 —————

Age (years) 43±19 18–92BMI (Kg/m2) 27.4±6.1 17.6–48.1TC (mg/dl) 196±41 105–341LDLc (mg/dl) 126±35 47–245HDLc (mg/dl) 48±15 21–93Triglycerides (mg/dl) 93 (63,146) 28–591Apo B (mg/dl) 92±26 46–178AIP 0.319±0.337 −0.320–1.450

Data are expressed as mean±standard deviation for parametric data or as median(25th and 75th percentiles) for non-parametric data.Abbreviations: BMI: bodymass index; TC: total cholesterol, LDLc: low-density lipoproteincholesterol, HDLc: high-density lipoprotein cholesterol; Apo B: Apolipoprotein B, AIP:atherogenic index of plasma=(log (triglycerides/HDL cholesterol)).

with levels higher than 116 mg/dl. Furthermore, these subjects showeda greater risk of CVD, as reflected by their higher levels of total choles-terol and LDL cholesterol and lower levels of HDL) than subjects in thefirst and second tertiles (data not shown).

Finally, we analysed diameter and pattern of LDL –both measuredby the PTGE and PGGE methods — according to triglyceride tertiles.We found a significant reduction in the diameter of LDL particles inpatients in the third tertile when compared with those in the firstand second tertiles (Fig. 3A). In line with this, pattern A was predom-inant among subjects in the first and second tertiles, with only six forPTGE and five for PGGE showing pattern B (0% and 9.7% in the firstand second tertiles, respectively, with PTGE and 0% and 8.1% in thefirst and second tertiles, respectively, with PGGE), while pattern Bwas predominant among subjects in the third tertile (highest triglyc-eride levels) (41.4% with PTGE and 46.6% with PGGE) (Fig. 3B).

4. Discussion

In the last decade, there has been increasing interest in characterizingand measuring LDL subfractions due to their involvement in the athero-sclerotic process. Among the different approaches currently employed,PGGE, preceded by ultracentrifugation and nuclear magnetic resonancespectroscopy, is awell-established andprecisemethod for the separationof LDL subfractions, though it is time-consuming and technically de-manding. As new methods are introduced (for example, PTGE), stan-dardization becomes increasingly important, as the principles on whicheach approach is based can vary, and there is little information to com-pare the various analytical techniques available. As we have mentionedin Material & Methods, the PTGE method developed by QuantimetrixLipoprint® system (Quantimetrix Corporation, Redondo Beach, CA,USA) allows the separation of LDL subfractions into 7 bands based pri-marily on LDL electrophoretic mobility. LDL subfractions 1 and 2 havebeen designated as large LDL, and subfractions 3 to 7 as sdLDL, with asubsequent reduction in LDL size as fraction number increases.

We analyzed LDL particle size in 177 samples using two different an-alytical methods: PTGE and PGGE. Bothmethods were highly correlated,though it must be pointed out that correlation is a measure of thestrength of a relation (or association) rather than agreement betweentwo factors. Due to the underlying statistical assumption of correlationcoefficients, interpretation of test accuracy using these values can bemis-leading, as a correlation coefficient depends on the range of values in thesample, and the correlation is likely to be stronger when the range iswide rather than narrow. Furthermore, it is difficult to assign a correla-tion coefficient threshold value for widespread acceptance of a test.

In this context, we focused on a diagnostic classification of patientLDL pattern type that goes beyond the quantitative data of particle size,as the latter parameter is often not comparable due to differences be-tween the methodologies employed. Our results indicate that 94.4% ofthe study population had the same LDL particle pattern, with only 10 pa-tients being misclassified. Despite this, there was very good agreementbetween the 2 methods, with a weighted kappa index of 0.81. Of the32 samples classified by PGGE as sdLDL particles, 6 were misclassifiedby PTGE as large LDL particles (81.3% concordance), and of the 145 sam-ples classified by PGGE as large, 4 were misclassified by PTGE as small(97.2% concordance). Consequently, this method of analysis appears topermit a simplified categorization of patients as displaying pattern A orpattern B, with good correlation and concordance with PGGE. To ourknowledge, only three studies have addressed this topic previously,and all three classified LDL subfractions as small, intermediate or largebased on different cut-off points. In accordance with our results, Hiranyet al. [13] reported good agreement between PTGE and PGGE methodsafter evaluating the data using kappa statistics (weighed K=0.78),with the former method showing an agreement of 92%, 33% and 77%concordance for classification of small, intermediate and large LDL sub-fractions, respectively, when compared with PGGE. Similarly, anotherstudy reported 100%, 64%, and 95% agreement for classification of

29

A B4

r=0.568, p<0.001

26

27

28

r = -0.617, p<0.001

0

1

2

3

24

25

LDL diameter by PTGE (nm)

LDL

diam

eter

by

PG

GE

(nm

)

-2

-1

25 26 27 28 24.5 25.5 26.5 27.5

Average LDL size (nm)

LDL

size

dife

renc

e (n

m)

Fig. 2. LDL size analyzed by PGGE plotted against LDL size measured by PTGE (A) and Bland-Altman plot of absolute differences between LDL size determined by PTGE and PGGE vsthe average of the two methods, with the mean of the differences (solid line) and mean±2SD limits (broken line) (B). Abbreviations: PTGE: polyacrylamide tube gel electropho-resis; PGGE: polyacrylamide gradient gel electrophoresis.

Table 3Correlation of triglycerides, atherogenic index and Apo B of plasma with LDL particlesize measured by polyacrylamide gel gradient electrophoresis and polyacrylamidetube gradient electrophoresis.

255C. Bañuls et al. / Clinica Chimica Acta 413 (2012) 251–257

small, intermediate and large LDL subfractions, respectively, althoughthe kappa index was not provided [12]. In a more recent systematic re-view, the abovementioned concordance percentages were re-examined,showing a noticeable change from 100% to 88% and from 95% to 24% forsmall and large LDL subfractions, respectively [11]. These differenceswere based on the assignment of subjects with an intermediate pattern.Finally, Ensign and coworkers [14] compared LDL pattern as determinedby different technologies – PGGE, ultracentrifugation, nuclear magneticresonance and PTGE – and demonstrated disparities among themethods, showing that complete agreementwith respect to LDL subclassphenotype occurred in only 8% of the samples analyzed, reporting amere41% agreement between PTGE and PGGE with respect to the classifica-tion of LDL pattern (38% for large LDL and 3% for sdLDL). The small sam-ple sizes of these studies [12,14] and/or their categorization of LDL inthree patterns (the intermediate pattern being most prone to misclassi-fication) could justify the variability in concordance indexes reported.Additionally, computer software for automated analysis of the bandswas not available at the time the studies were published; therefore,each systemwould have had adifferent cut-off point for the classificationof each pattern. Currently, the automated system of PTGE supplied byQuantimetrix (Lipoware software) allows a standardized analysis ofLDL subfraction. In our Lipid Research Unit, we have observed that thePTGEmethod is precise and compares favourablywith the PGGEmethodwhen employed for diagnostic classification of LDL particle type. Thus,we would affirm that both analytical methods can be used interchange-ably. Additional advantages of PTGEwith respect to PGGE is that it is fas-ter and easier to performwhen assessing large numbers of samples, andtherefore could be of great practicality in clinical practice. However, PTGEis a relatively expensive procedure and may not be suitable for all pa-tients. As a consequence, its use should be limited to patients in whomthe determination of LDL particle size is likely to provide a real assess-ment of CAD risk.

Table 2Comparability of polyacrylamide tube gel electrophoresis and polyacrylamide gradientgel electrophoresis as methods of LDL subfraction determination.

PGGE Numberof cases

PTGE Concordance(%)*

Pattern A Pattern B

(LDL size≥26.8 nm)

(LDL sizeb26.8 nm)

Pattern A 145 141 4 97.2(LDL size ≥25.5 nm)Pattern B 32 6 26 81.3(LDL sizeb25.5 nm)

*weighed kappa of 0.81.Abbreviations: PTGE: polyacrylamide tube gel electrophoresis; PGGE: polyacrylamidegradient gel electrophoresis.

An increased proportion of sdLDL particles has been associated withmarked alterations in serum lipoprotein and lipid levels, in particular el-evated triglyceride concentrations. In accordance with that reported byother studies [12,13,20,23,24], we found that triglyceride and Apo Bconcentration –a indicator of the total number of LDL particles [25] –and AIP – considered a surrogate of sdLDL particle size [26]–were close-ly associated with LDL particle size. By dividing triglyceride levels intotertiles (firstb68mg/dl, second 68–116 mg/dl and thirdN116 mg/dl),we were able to observe that the correlation of triglyceride and AIPwith LDL particle size disappeared in the lower (first and second) ter-tiles, showing a good correlation, and one which was even better thanin the general population, when only subjects in the third tertile wereconsidered. Thus, triglyceride levels did not predict the LDL particlesize of subjects in the first and second tertiles. Other lipoproteins mayhave been responsible for the variation of LDL particle size in our popu-lation, although this parameter was relatively constant, since only six ofthe 119 subjects was classified as pattern B (5.0%). Thus, Apo B showeda good correlation with LDL size (measured by PTGE) in all tertiles,which is in accordance with the results of a recent study that comparedthe PTGE classification method with measurement of Apo B and directsdLDL [27]. In contrast, we detected a pattern that was predominantlyof the B type (41.4%) and a more atherogenic lipid profile among pa-tients in the third tertile that was associated with high total and LDLcholesterol (borderline high-risk), Apo B levels and AIP and low HDLcholesterol levels. In this way, our results suggest that determiningLDL subfractions in patientswith higher-than-normal triglyceride levels

PTGE PGGE

r* p r* p

Total triglycerides (mg/dl) −0.705 b0.001 −0.484 b0.0011st tertile 0.086 0.522 0.129 0.3382nd tertile −0.123 0.339 −0.028 0.8303rd tertile −0.676 b0.001 −0.527 b0.001

AIP (log (triglycerides/HDLc) −0.607 b0.001 −0.435 b0.0011st tertile 0.055 0.686 0.094 0.4872nd tertile −0.123 0.342 0.088 0.4963rd tertile −0.636 b0.001 −0.552 b0.001

Apo B (mg/dl) −0.532 b0.001 −0.244 0.0011st tertile −0.305 0.021 0.051 0.7052nd tertile −0.565 b0.001 −0.321 0.0123rd tertile −0.361 0.005 0.060 0.652

*Pearson correlation. 1st tertile included 57 subjects with triglyceride levels b68 mg/dl,2nd tertile included 62 subjects with triglyceride levels between 68 mg/dl and 116 mg/dland 3 rd tertile included 58 subjects with triglyceride levels N116 mg/dl.

PTGE27.5

A BPGGE

b

aa

26.7

26.9

27.1

27.3

aa

25.9

26.1

26.3

26.5

LDL

diam

eter

(nm

)

LDL

diam

eter

(nm

)

26.5

Triglycerides (mg/dl)

Pattern A

b

25.5

25.7

Triglycerides (mg/dl)

414

100 90.358.6

50

100

% o

f cas

es b

y P

TG

E

% o

f cas

es b

y P

GG

E

Pattern B

Pattern A

Pattern B

100 91.1

53.4

50

100

9.7

41.4

0 8.1

66

0

<68 68-116 >116 <68 68-116 >116

Triglycerides (mg/dl) Triglycerides (mg/dl)<68 68-116 >116 <68 68-116 >116

Fig. 3. Distribution of LDL diameter and pattern types measured by PTGE (A) and PGGE (B) according to triglyceride tertiles. Data are expressed as mean±standard error of 57, 62and 58 subjects for first, second and third tertile, respectively. Unlike superscript letters were significantly different when different tertiles were compared using one-way ANOVAand a Student-Newman-Keuls test as post hoc. Abbreviations: PTGE: polyacrylamide tube gel electrophoresis; PGGE: polyacrylamide gradient gel electrophoresis.

256 C. Bañuls et al. / Clinica Chimica Acta 413 (2012) 251–257

(N150 mg/dl) could be an important tool for assessing CAD risk, as thisrisk increases when values exceed 116 mg/dl.

Considered as awhole, our results show that PTGE correlates favour-ably and is in good agreement with PGGE when used for the diagnosticclassification of LDL pattern type. This evidence has led us to employboth analyticalmethods interchangeably in our Lipid ResearchUnit. Ad-ditionally, the determination of LDL particle size would seem to be anappropriate analytical procedure for estimating CAD risk in patientswith borderline high-risk levels of total and LDL cholesterol and withtriglyceride levels higher than 116 mg/dl.

List of abbreviations

AIP atherogenic index of plasmaApo B apolipoprotein BCAD coronary artery diseaseCHD coronary heart diseaseCVD cardiovascular diseaseFDA Food and Drug AdministrationK kappa indexHDL high-density lipoproteinLDL low-density lipoproteinPGGE polyacrylamide gradient gel electrophoresisPTGE polyacrylamide tube gel electrophoresisRf retardation factorsdLDL small dense low-density lipoproteinVLDL very-low density lipoprotein

Conflict of interest

The authors claim no conflict of interest associated with themanuscript.

Acknowledgements

Sources of financial support: The study has been supported bygrants PS09/01025 and PI10/1195 from Fondo de Investigación Sani-taria (FIS) and AP-192/11 from Regional Ministry of Health of Valen-cian Community. M Rocha is a recipient of Miguel Servet contractfrom FIS (CP10/00360). VM Víctor is a recipient of Regional Ministryof Health of Valencian Community and Carlos III Health Institute con-tract (CES10/030).

We kindly thank B Normanly and I Soria-Cuenca for their contri-bution to the present study.

References

[1] Singh BK, Mehta JL. Management of dyslipidemia in the primary prevention ofcoronary heart disease. Curr Opin Cardiol 2002;17:503–11.

[2] Austin MA, Breslow JL, Hennekens CH, Buring JE, Willett WC, Krauss RM. Low-density lipoprotein subclass patterns and risk of myocardial infarction. JAMA1988;260:1917–21.

[3] Krauss RM, Siri PW. Metabolic abnormalities: triglyceride and low-density lipo-protein. Endocrinol Metab Clin North Am 2004;33:405–15.

[4] Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol inAdults. Executive Summary of The Third Report of The National Cholesterol EducationProgram (NCEP) Expert Panel on Detection, Evaluation, And Treatment of High BloodCholesterol In Adults (Adult Treatment Panel III). JAMA 2001;285:48–97.

[5] Rizzo M, Berneis K. Low-density lipoprotein size and cardiovascular prevention.Eur J Intern Med 2006;17:77–80.

[6] Otvos JD, Mora S, Shalaurova I, Greenland P, Mackey RH, Goff Jr DC. Clinical impli-cations of discordance between low-density lipoprotein cholesterol and particlenumber. J Clin Lipidol 2011;5:105–13.

[7] Hirano T, Ito Y, Yoshino G. Measurement of small dense low-density lipoproteinparticles. J Atheroscler Thromb 2005;12:67–72.

[8] Griffin BA, Caslake MJ, Yip B, Tait GW, Packard CJ, Shepherd J. Rapid isolation oflow density lipoprotein (LDL) subfractions from plasma by density gradient ultra-centrifugation. Atherosclerosis 1990;83:59–67.

[9] Otvos JD, Jeyarajah EJ, Bennett DW. Quantification of plasma lipoproteins by pro-ton nuclear magnetic resonance spectroscopy. Clin Chem 1991;37:377–86.

257C. Bañuls et al. / Clinica Chimica Acta 413 (2012) 251–257

[10] Austin MA, Hokanson JE, Brunzell JD. Characterization of lowdensity lipoproteinsubclasses: methodological approaches and clinical relevance. Curr Opin Lipidol1994;5:395–403.

[11] Chung M, Lichtenstein AH, Ip S, Lau J, Balk EM. Comparability of methods for LDLsubfraction determination: A systematic review. Atherosclerosis 2009;205:342–8.

[12] Hoefner DM, Hodel SD, O'Brien JF, et al. Development of a rapid, quantitativemethod for LDL subfractionation with use of the Quantimetrix Lipoprint LDL Sys-tem. Clin Chem 2001;47:266–74.

[13] Hirany SV, Othman Y, Kutscher P, Rainwater DL, Jialal I, Devaraj S. Comparison oflow-density lipoprotein size by polyacrylamide tube gel electrophoresis and poly-acrylamide gradient gel electrophoresis. Am J Clin Pathol 2003;119:439–45.

[14] EnsignW, Hill N, Heward CB. Disparate LDL phenotypic classification among 4 dif-ferent methods assessing LDL particle characteristics. Clin Chem 2006;52:1722–7.

[15] Allain CC, Poon LS, Chan CS, RichmondW, Fu PC. Enzymatic determination of totalserum cholesterol. Clin Chem 1974;20:470–5.

[16] Bucolo G, David H. Quantitative determination of serum triglycerides by the use ofenzymes. Clin Chem 1973;19:476–82.

[17] Sugiuchi H, Uji Y, Okabe H, et al. Direct measurement of high-density lipoproteincholesterol in serum with polyethylene glycol-modified enzymed and sulfatesalphacyclodextrin. Clin Chem 1995;41:717–23.

[18] FriedewaldWT, LevyRI, FredricksonDS. Estimation of the concentration of low-densitylipoprotein cholesterol in plasmawithout use of preparative ultracentrifuge. Clin Chem1972;18:499–502.

[19] Nichols AV, Krauss RM, Musliner TA. Nondenaturing polyacrylamide gradient gelelectrophoresis. Methods Enzymol 1986;128:417–31.

[20] Kazumi T, Kawaguchi A, Hozumi T, et al. Low density lipoprotein particle diameterin young, nonobese, normolipidemic Japanese men. Atherosclerosis 1999;142:113–9.

[21] Altman DG. Some common problems in medical research. In: Altman DG, editor.Practical statistics for medical research. New York: Chapman and Hall; 1991.p. 396–437.

[22] Bland JM, Altman DG. Statistical methods for assessing agreement between twomethods of clinical measurement. Lancet 1986;1:307–10.

[23] Lamarche B, Tchernof A, Moorjani S, et al. Small, dense low-density lipoproteinparticles as a predictor of the risk of ischemic heart disease in men. Prospectiveresults from the Québec Cardiovascular Study. Circulation 1997;95:69–75.

[24] Gardner CD, Fortmann SP, Krauss RM. Association of small low-density lipopro-tein particles with the incidence of coronary artery disease in men and women.JAMA 1996;276:875–81.

[25] Contois JH, McConnell JP, Sethi AA, et al. Apolipoprotein B and cardiovascular dis-ease risk: Position statement from the AACC Lipoproteins and Vascular DiseasesDivision Working Group on Best Practices. Clin Chem 2009;55:407–19.

[26] Dobiásová M, Frohlich J. The plasma parameter log (TG/HDL-C) as an atherogenicindex: correlation with lipoprotein particle size and esterification rate in apoB-lipoprotein-depleted plasma (FER(HDL)). Clin Biochem 2001;34:583–8.

[27] Vandermeersch A, Ameye S, Puype D, Petitjean D, De Buyzere M, Langlois MR. Es-timation of the low-density lipoprotein (LDL) subclass phenotype using a direct,automated assay of small dense LDL-cholesterol without sample pretreatment.Clin Chim Acta 2010;411:1361–6.