Journal of Cardiology (2008) 52, 118—126

ORIGINAL ARTICLE

Association of low-density lipoprotein particlesize and ratio of different lipoproteins andapolipoproteins with coronary heart disease

Santanu Biswas (MSc)a, Pradip K. Ghoshal (MD)b,Sankar C. Mandal (MD, DM)b, Nripendranath Mandal (MSc, PhD)a,∗

a Immunotechnology Section, Bose Institute, P-1/12 CIT Scheme VIIM,Kolkata 700054, West Bengal, Indiab Departments of Cardiology, N.R.S. Medical College & Hospital,138 A.J.C. Bose Road, Kolkata 700014, India

Received 5 March 2008; received in revised form 2 June 2008; accepted 16 June 2008Available online 29 July 2008

KEYWORDSCoronary heart disease;Apolipoproteins;LDL particle;ROC curve

SummaryBackground: Worldwide coronary heart disease (CHD) is estimated to be the leadingcause of death. Current knowledge about prevention of CHD is mainly derived fromdeveloped countries. Therefore, this study aimed to find out the association of CHDwith ratios of different lipoproteins and apolipoproteins, LDL particle size, as wellas different traditional risk factors in Asian Indian population in Eastern part ofIndia.Methods: Case—control study of 100 patients with CHD and 98 healthy controls wereage and sex matched. After clinical evaluation, blood samples were collected forbiochemical assays.Results: Multivariate logistic regression analysis found apoB (OR 2.96; 95% CI1.02—8.54), apoB/HDL-c (OR 4.14; 95% CI 1.33—12.83), nonHDL-c (OR 5.41; 95% CI2.08—14.10), apoB/apoAI (OR 6.64; 95% CI 2.37—18.57), and LDL particle size (9.59;95% CI 2.92—31.54) were independently associated with CHD. Area under the ROCcurves derived from the model (AUROC 0.947; 95% CI 0.916—0.977) are significantlyhigher than any other variables.Conclusions: Findings from the multivariate analysis, apoB, apoB/HDL-c, nonHDL-c,

apoB/apoAI, and LDL particle size are potent indicators and useful for diagnosis ofpredisposed CHD.© 2008 Japanese College of Cardiology. Published by Elsevier Ireland Ltd. All rightsreserved.

∗ Corresponding author. Tel.: +91 33 2355 9219/9416/9544x497; fax: +91 33 2355 3886.E-mail addresses: mandaln@rediffmail.com, nripen@boseinst.ernet.in (N. Mandal).

0914-5087/$ — see front matter © 2008 Japanese College of Cardiology. Published by Elsevier Ireland Ltd. All rights reserved.doi:10.1016/j.jjcc.2008.06.005

L

I

CpChppUCtirpmprot[t[c(bpao

tsikddgteovw

adatiHmaaciaAI

M

Pc

ApasiysbtwcteedopOdittIoippqppattwotubia

Q

Qt

DL particle size and lipoproteins/apo in CHD

ntroduction

oronary heart diseases (CHD) account for a largeroportion of all deaths and disability worldwide.oronary artery disease (CAD) [1] and ischemiceart disease [2] are the synonyms of CHD. Therevalence of CHD is known to be high among peo-le of Asian Indian origin. Indians settled in theSA have a four fold higher prevalence of CHD thanaucasian Americans and six fold higher hospitaliza-ions than Chinese Americans [3]. Rates are risingn India, and by 2015 CHD has been predicted toank first among the causes of death in the Indianopulation [4]. The high tendency to develop pre-ature and accelerated CHD in Asian Indians is onlyartially explained by the presence of conventionalisk factors such as insulin resistance and abdominalbesity [5], etc. Among the conventional risk fac-ors, elevated levels of serum total cholesterol (TC)6], low density lipoprotein cholesterol (LDL-c) [7],riglycerides (TG) [8,9], apolipoprotein B (apoB)10], and lower levels of high density lipoproteinholesterol (HDL-c) [11,12] and apolipoprotein AIapoAI) [13] are the well known and establishediomarker in CHD. Small dense LDL is considered toromote atherosclerosis in CHD because of its lowffinity for LDL receptors [14] and susceptibility toxidative modification [15].

Since the number of deaths in the Indian popula-ion due to CHD have drastically increased, it is noturprising that the medical fraternity is now focus-ng more on finding preventive measures. Currentnowledge about the prevention of CHD and car-iovascular disease is mainly derived from studiesone in populations of European and American ori-in. As far as the eastern part of India is concerned,here was no study of different apolipoproteins lev-ls, assessment of LDL particle size, and some ratiosf lipoproteins and apolipoproteins with the con-entional risk factors in blood, which are connectedith CHD.We have, therefore, attempted to find out the

ssociation of CHD with LDL particle size, ratios ofifferent lipoproteins and apolipoproteins, as wells different traditional risk factors and the diagnos-ic performance of this newly proposed biomarkern CHD compared against known lipid markers.ence this investigation was undertaken to deter-ine whether the ratios of different lipoproteins

nd apolipoproteins are associated with CHD andlso whether apoB/apoAI ratio is better than theholesterol ratios to predict CHD. So it is very

mportant to study the newly arisen biomarkerlong with the conventional risk factors amongsian Indian population in the eastern part of

ndia.

tlatc

119

ethods

opulation description and sampleollection

ll subjects are Indian adults (1) CHD group: 100atients (86 male, 14 female) with typical anginand electrocardiographic study, tread mill test,tress echo and echocardiographic evidence ofschemia or infarction, aged between 45 and 65ears old. (2) Control group: 98 healthy age andex-matched subjects (86 male, 12 female) agedetween 45 and 65 years. The controls comprisedhe spouses, neighbors, and people from sameork place of the patients, with the same socio-ultural background, in whom the clinical history,he objective search for signals of CHD, and thelectrocardiographic as well as echocardiographicxamination did not suggest the presence of thatisease. All patients and controls with ancestralrigin were from the eastern part of India. Theresent study was conducted during the period ofctober 2006—February 2007 at the out-patientsepartment and also with in-patients of NRS Med-cal College & Hospital, Kolkata. The subjects ofhe present study were part of a health examina-ion between the Immunotechnology Section, Bosenstitute, Kolkata and the Department of Cardiol-gy, NRS Medical College & Hospital, Kolkata. Thenstitutional ethical committee approved the studyrotocol. Informed consent was obtained from thearticipants. The cardiologist completed a clinicaluestionnaire for each subject. Venous blood sam-les (5 ml) were collected in the morning at theoint of medical check-up into a sterile tube after12-h overnight fast and also before the patient

ook any lipid-lowering drugs. Any patients or con-rols found to have taken any lipid-lowering drugsere excluded from the study. Serum was harvestedn the day of sample collection by centrifuga-ion at 3000 rpm for 10 min at room temperaturesing tabletop centrifuge (Remi Pvt. Ltd., Mum-ai, India). Subsequently, the serum was dividednto aliquots for determination of lipids, glucose,poA1, apoB, and apoE analysis.

uestionnaires and clinical characteristics

uestionnaires were distributed at the time ofhe medical check-up. Participants were ques-

ioned about smoking and about any use ofipid-lowering medication. They were questionedbout the amount of alcoholic beverage (coun-ry liquor) drunk per day as well as number ofigarettes, or ‘‘biri’’ (a type of local cigarette)

120 S. Biswas et al.

Table 1 Characteristic of Asian Indian population in Eastern part of India in the case and control group

Variable Case (N = 100) Control (N = 98) p-Value a

Age, year (mean, S.D.) 54.80 (8.60) 55.54 (9.73) Matched

Sex (n, %) MatchedMale 86 (86%) 86 (87.75%)Female 14 (14%) 12 (12.25%)

Smoking 0.000Yes 63 (63%) 27 (27.55%)No 27 (27%) 71 (72.45%)

Hypertension 0.000Yes 66 (66%) 13 (13.27%)No 34 (34%) 85 (86.73%)

Alcohol 0.000Yes 27 (27%) 12 (12.25%)No 73 (73%) 86 (87.75%)

con

rPLpGvLL

Waist circumference (cm) 88.75 (7.15)a Mann—Whitney test for comparisons between the case and

smoked per day. Information about personal andfamily history of cardiovascular disease, and riskfactors (hypertension) was obtained. Waist circum-ference was measured just above the naval overlight clothing, using unstreatched tape meter, with-out any pressure to body surface, and was recordedto the nearest 0.1 cm.

Blood biochemistry

Glucose was determined by using GOD-POD reagent(Merck Ltd., Mumbai, India). TC was determinedenzymatically using CHOD/POD-Phosphotungstate

atdas

Table 2 The demographic information in CHD group and c

Variables CHD (n = 100)

Glucose (mg/dl) 90.91 (29.89)Total cholesterol (mg/dl) 191.37 (36.42)LDL-c (mg/dl) 124.45 (33.28)HDL-c (mg/dl) 29.03 (6.33)NonHDL-c (mg/dl) 162.17 (36.09)TG (mg/dl) 188.61 (52.38)apoA1 (mg/dl) 119.69 (21.90)apoB (mg/dl) 114.74 (16.14)apoE (mg/dl) 4.14 (0.72)LDL-c/HDL-c 4.44 (1.42)TC/HDL-c 6.80 (1.73)TG/HDL-c 6.78 (2.90)apoB/ApoA1 0.99 (0.24)apoB/HDL-c 4.13 (1.02)LDL-c/apoB 1.09 (0.29)HDL-c/apoAI 0.25 (0.07)NonHDL-c/HDL-c 5.80 (1.73)LDL particle size (nm) 24.60 (1.02)

a Non paired Student’s t-test for comparisons between the mean

85.42 (5.17) 0.000

trol group.

eagent. HDL-c was determined with CHOD/POD-hosphotungstate reagent (Accurex Biomedical Pvt.td., Mumbai, India), after precipitation withhosphotungstic acid. TG was determined usingOP-POD reagent (Accurex Biomedical). Besidesalues of LDL-c was estimated using the formulaeDL-c = TC − (HDL-c + FTG/5) [16]. Determination ofDL particle size is based on the method of Krauss

nd Burke using gradient polyacrylamide gel elec-rophoresis [17]. Apolipoprotein AI, B, and E wereetermined by immunoturbidimetry method usinguto N ‘‘DAIICHI’’ reagent and for standard mea-urement using Apo auto N ‘‘DAIICHI’’ calibrator.

ontrol groups (mean, S.D.)

Control (n = 98) p-Valuea

81.06 (18.46) 0.006161.11 (32.11) 0.00099.97 (28.22) 0.00034.18 (7.66) 0.000

126.93 (31.34) 0.000134.77 (48.04) 0.000129.79 (22.49) 0.00289.06 (15.88) 0.0004.11 (0.70) 0.7693.10 (1.45) 0.0004.94 (1.71) 0.0004.19 (1.98) 0.0000.69 (0.13) 0.0002.73 (0.82) 0.0001.15(0.39) 0.2130.27 (0.07) 0.0563.94 (1.71) 0.000

25.59 (0.67) 0.000

value of case and control groups.

LDL particle size and lipoproteins/apo in CHD 121

Table 3 Area under the receiver operating characteristic curve (AUROC) for different variables

Variables AUROC Standard error p-Value 95% CI

Glucose 0.591 0.042 0.029 0.508—0.673TC 0.767 0.035 0.000 0.699—0.836LDL-c 0.743 0.036 0.000 0.673—0.814HDL-c 0.721 0.037 0.000 0.577—0.864NonHDL-c 0.800 0.033 0.000 0.735—0.865TG 0.798 0.032 0.000 0.737—0.860apoAI 0.623 0.040 0.003 0.545—0.700apoB 0.885 0.023 0.000 0.840—0.931apoE 0.519 0.042 0.644 0.437—0.601LDL-c/HDL-c 0.806 0.032 0.000 0.743—0.868TC/HDL-c 0.835 0.030 0.000 0.777—0.893TG/HDL-c 0.835 0.029 0.000 0.777—0.893NonHDL-c/HDL-c 0.835 0.030 0.000 0.777—0.893apoB/apoAI 0.882 0.023 0.000 0.838—0.927apoB/HDL-c 0.882 0.025 0.000 0.833—0.931LDL-c/apoB 0.504 0.042 0.914 0.423—0.586

08

Fs

S

Awawcs

wuDap(w

HDL-c/apoAI 0.599 0.04LDL particle size 0.860 0.02

or these measurements we used Microlab 2000emi-auto analyzer (Merck Ltd.).

tatistical analysis

fter the completion of each experiment, the dataere recorded on pre-designed proforma and man-

ged with Microsoft Excel software. Data entryas double checked for any human error. All cal-ulations were performed using SPSS version 10.0oftware package for windows and MedCalc soft-

fcmo

Table 4 Optimal cut-off value, sensitivity, specificity, and pvalues (NPV) of different variables

Variables Cut-off Sensitivit

WC (cm) 85 0.680Glucose (mg/dl) 100 0.313TC (mg/dl) 175 0.724LDL-c (mg/dl) 108 0.694HDL-c (mg/dl) 30 0.626NonHDL-c (mg/dl) 140 0.775TG (mg/dl) 150 0.786apoAI (mg/dl) 121 0.500apoB (mg/dl) 100 0.830apoE (mg/dl) 3.95 0.490LDL-c/HDL-c 3.40 0.765TC/HDL-c 5.30 0.806TG/HDL-c 4.91 0.816NonHDL-c/HDL-c 4.42 0.806apoB/apoA1 0.80 0.790apoB/HDL-c 3.07 0.879LDL-c/ApoB 1.04 0.520HDL-c/apoAI 0.24 0.566LDL particle size (nm) 25.5 0.663

0.016 0.520—0.6780.000 0.806—0.914

are were used for comparison between the areander the receiver operating characteristic curve.ata are presented as means (S.D.) and percent-ge of the population studied. Student’s t-test (forarametric variables) and Mann—Whitney U-testfor non-parametric variables) were used where itas applicable to estimate the significance of dif-

erence between two groups. Receivers operatingharacteristic (ROC) curves were used to deter-ine the optimal cut-off values of these ratios

f lipoprotein and apolipoproteins. The points of

ositive predictive values (PPV) and negative predictive

y Specificity PPV NPV

0.531 0.596 0.6200.877 0.720 0.5580.755 0.747 0.7320.684 0.687 0.6910.724 0.697 0.6570.765 0.767 0.7730.673 0.706 0.7580.673 0.610 0.5690.734 0.761 0.8080.479 0.490 0.4790.704 0.721 0.7500.724 0.745 0.7890.745 0.762 0.8020.776 0.782 0.8000.735 0.752 0.7740.694 0.743 0.8500.469 0.495 0.4940.574 0.574 0.5660.910 0.880 0.739

122

Table 5 Univariate logistic regression analysis of dif-ferent markers in CHD

Variables Odds ratio 95% CI p-Value

Smoking 4.47 2.45—8.16 0.000WC 2.40 1.34—4.28 0.003Hypertension 12.69 6.20—25.95 0.000Alcohol intake 2.65 1.25—5.60 0.009Glucose 3.26 1.56—6.83 0.001TC 8.10 4.27—15.36 0.000LDL-c 4.89 2.67—8.96 0.000HDL-c 4.40 2.41—8.04 0.000NonHDL-c 11.26 5.78—21.92 0.000TG 7.56 3.98—14.35 0.000apoAI 2.06 1.15—3.66 0.013apoB 13.52 6.79—26.90 0.000apoE 0.880 0.50—1.54 0.669LDL-c/HDL-c 7.75 4.10—14.67 0.000TC/HDL-c 10.93 5.60—21.34 0.000TG/HDL-c 12.97 6.54—25.71 0.000NonHDL-c/HDL-c 14.36 7.20—28.63 0.000apoB/apoA1 10.41 5.39—20.11 0.000apoB/HDL-c 16.43 7.83—34.47 0.000LDL-c/apoB 0.960 0.541—1.682 0.886

estwt9aAeo

R

W1fa[(hcwThti

HDL-c/apoAI 1.73 0.981—3.50 0.054LDL particle size 20.85 9.32—46.63 0.000

convergence of sensitivity and specificity deter-

mined the optimal cut-off points for these riskfactors. Then all the continuous variables weredichotomized. Independent indicators for the pres-

Figure 1 Receiver-operating characteristic curve for theprediction model of coronary heart disease. The ROCcurve was drawn from the equation that was calcu-lated by logistic regression and is as follows: disease [yesor no] = 1.085 apoB + 1.421 apoB/HDL-c + 1.689 nonHDL-c + 1.849 apoB/apoAI + 2.262 LDL particle size − 11.702.

tltLcthtcrhavovcstHa

lacaahts

S. Biswas et al.

nce of CHD were also selected by a forwardtepwise conditional logistic regression analysis. Inhis method, the lipid and non-lipid parametersere selected into the model in the order of statis-

ical significance. For all odds ratios, we calculated5% confidence intervals (CIs) of each variable andssociation study was done with Pearson �2-test.ssociation of variables having p < 0.05 was consid-red as statistical significance. All statistical testsf hypothesis are two sided.

esults

e studied a total of 198 subjects, of which00 patients had CHD [86 male (86%) and 14emale (14%)], mean (S.D.) age 54.80 (8.60) yearsnd 98 age and sex-matched healthy controls86 male (87.75%) and 12 female (12.25%)] meanS.D.) age 55.54 (9.73) (Table 1). Smoking, alco-ol consumption, hypertension and increased waistircumference were highly prevalent in the groupith CHD as compared to control group (Table 1).he mean value of TC and LDL-c were significantlyigher in CHD patients compared with controls, buthe mean value of HDL-c was significantly lowern CHD patients than controls. We also observedhat the mean value of TG and blood glucoseevels was higher in CHD patient than the con-rol group (Table 2). Whereas the mean ratio ofDL-c/HDL-c, TC/HDL-c, TG/HDL-c and nonHDL-/HDL-c were higher among CHD patients comparedo controls and all the results were statisticallyighly significant (Table 2). Among the apolipopro-eins, mean value of apoB was higher in patientsompared to controls and value of apoAI was justeverse of it and both the values were statisticallyighly significant, on the other hand mean value ofpoE was similar in patients and controls and thealue was statistically insignificant. But the ratiof apoB/apoAI was highly significant and the meanalue was higher in CHD patients compared to theontrol group (Table 2). Among the studied ratios,ome are, e.g. apo B/HDL-c shown to be statis-ically significant. On the contrary LDL-c/apo B,DL-c/apoA1 ratios are lower in CHD patients butre statistically insignificant.

Diagnostic implications of ratios of differentipoproteins and apolipoproteins against the lipidsnd apolipoprotein markers were assessed by ROCurves analysis. The analysis demonstrated that

poB (AUROC 0.885), apoB/apoAI (AUROC 0.882),nd apoB/HDL-c (AUROC 0.882) were shown toave superior discriminative ability for CHD againsthe other parameters (Table 3). There was notatistically significant difference found in AUROC

LDL particle size and lipoproteins/apo in CHD 123

Table 6 Multivariate logistic regression analysis

Independent variables Regression coefficient Odds ratio 95% CI p-Value

apoB 1.085 2.96 1.02—8.54 0.045apB/HDL-c 1.421 4.14 1.33—12.83 0.014NonHDL-c 1.689 5.41 2.08—14.10 0.000apoB/apoAI 1.894 6.64 2.37—18.57 0.000

bcd(

hcTratLCsacthBatt

veroptapL(fChaL

D

Tnw

i[totrap

wriLeantchgrlncd

lcsaealMiotwsCs

LDL particle size 2.262

etween these three variables. In Table 4 optimalut-off value, sensitivity, specificity, positive pre-ictive value (PPV) and negative predictive valueNPV) of different parameters are shown.

Among the categorical variables, hypertensionad 12 times higher odds ratio in patients thanontrols and increased risk for CHD. The data inable 5 indicate the risk of CHD of conventionalisk factors along with the different lipoproteinnd apolipoprotein ratios and LDL particle size. Inhe present study it was observed that those withDL particle size less than 25.5 nm have risks ofHD 20 times higher than those with LDL particleize more than 25.5 nm. The risk for the ratio ofpoB/HDL-c, nonHDL-c/HDL-c, TG/HDL-c, TC/HDL-, apoB/apoAI and LDL-c/HDL-c are 16 times, 14imes, 12 times, 10 times, 10 times and 7 timesigher in disease than control group, respectively.ut the odds ratios of LDL-c/apoB and HDL-c/apoAIre not significant. Among the conventional risk fac-ors the risk of apoB and nonHDL-c are 13 and 11imes higher in CHD patients than controls.

Regression models were used to predict thealue of a response variable using the dichotomizedxplanatory variables. From the univariate logisticegression analysis 15 potential clinical predictorsf CHD were evaluated (Table 5). But in the finalredictions model (Fig. 1) only five of these fac-ors: the presence of nonHDL-c (OR 5.41; p = 0.001),poB (OR 2.96; p = 0.045), apoB/apoAI (OR 6.64;= 0.000), apoB/HDL-c (OR 4.41; p = 0.014) andDL particle size (OR 9.59; p = 0.000) were foundTable 6). Area under the ROC curves derivedrom the multivariate model (AUROC 0.947; 95%I 0.916—0.977) were statistically significantlyigher than nonHDL-c (p < 0.001), apoB (p = 0.019),poB/apoAI (p = 0.009), apoB/HDL-c (p = 0.009) andDL particle size (p = 0.001).

iscussion

he exact etiology of CHD is unknown; a largeumber of risk factors are known to be associatedith CHD. Obesity [18], hypertension [19], smok-

am1m

9.59 2.92—31.54 0.000

ng [20,21], family history [22], diabetes mellitus23], and plasma lipoprotein abnormalities [23] arehe conventional risk factors of this disease. But inur knowledge, limited studies have been under-aken to investigate the relationship between theatios of different lipoproteins and apolipoproteinsnd LDL particle size with CHD in the Asian Indianopulation in eastern part of India.

The aim of this study was to identify patientsith early CHD associated with the conventional

isk factors and some of the new risk biochem-cal markers, including apolipoprotein variables,DL particle size, in addition to ratios of differ-nt lipoproteins and apolipoproteins. These ratiosre chosen because human physiological systemseed all the lipoproteins as well as apolipopro-eins but it has certain limits. When these limitsross, the systems face a lot of problems. So theuman physiological systems need to balance all theood and bad things. However, here we consideredatios of different lipoproteins and apolipoproteinsevels such as LDL-c/HDL-c, TC/HDL-c, TG/HDL-c,onHDL-c/HDL-c, apoB/apoAI, apoB/HDL-c, LDL-/apoB and HDL-c/apoAI are important factors foriagnosis of premature CHD.

The relevant findings in early CHD were as fol-ows: hypertension, waist circumference, alcoholonsumption and smoking were statistically highlyignificant with the CHD. Abnormal levels of lipidsnd apolipoproteins characterize the patients witharly CHD who had a type of dyslipidemia withn increase in atherogenic lipoproteins and apoB,ower levels of HDL-c and apoAI. According toasunaga et al. moderate drinking reduces the

ncidence of cardiovascular events [24], but inur study we found that 27% and 12.25% amongotal number of patients and controls, respectively,ere drinking country liquor per day, and the data

how alcohol intake is significantly associated withHD (Table 1). A study by Stamler et al. found atrong association between serum cholesterol, CAD

nd cardiovascular death [6]. Gandhi reported thatean serum TC level was 160 mg/dl in males and

50 mg/dl in females [25]. Our study shows thatean TC in patients is 191.37 mg/dl and the con-

oaw[plcralHhcHlwCn

tttanimAf(cnadruodTc(faftbdHr

et

124

trol group is 161.11 mg/dl. LDL-c is the strongestpredictor of CHD and some studies show that thedisease is closely correlated with high concentra-tions of TC and LDL-c [7]. Natio reported a highincidence of atherosclerosis and CHD in subjectswith LDL-c above 130 mg/dl [26]. In our study wehave observed that an elevated level of LDL-c isassociated with CHD. The high concentrations ofTC and LDL-c in patients means that they are 8and 4 times more at risk than those with normalconcentrations, respectively. Numerous epidemi-ological studies from North America and Europehave conclusively demonstrated that high levels ofHDL-c protect against CHD [11]. According to Stainand Myers HDL-c below 35 mg/dl was associatedwith atherosclerosis and CHD [27]. Another groupshowed that low HDL-c or smoking is more likelyto result in myocardial infarction [28]. But in ourobservation we have found that below 30 mg/dlHDL-c concentrations is a risk for CHD in the easternpart of the Indian population. According to Wiklundet al. TG level above 200 mg/dl is a risk for CHD [8].In our study, hypertriglyceridemia alone was notobserved in the group with early CHD. However, theratio of LDL-c to HDL-c or TC to HDL-c is acceptedas an extremely important indicator of atherogen-esis [11]. In consideration of these, we also studiedthe relationship of these ratios with CHD, and foundthat these ratios are highly significantly associ-ated with CHD, as found by other groups [29]. TheAmerican Diabetes Association (64th Scientific Ses-sion, 2004, Orlando, FL, USA) proposed apoB/HDL-cratio as the predictor of atherosclerotic disease.We observed that the ratio of apoB/HDL-c andHDL-c is strongly associated with CHD risk amongAsian Indian populations in the eastern part ofIndia. According to Maruyama et al. concentrationof small dense LDL is positively associated withTG/HDL-c ratio [30]. In our study, we observed thatthe ratio of TG to HDL-c is higher in CHD patientscompared to the control group. Several studies haverelated the higher level of apoB and lower level ofapoAI to the early occurrence of CHD. We observedthat apoAI concentration is lower and apoB con-centration is higher in the CHD group than controlsas found in other studies [13]. A large number ofepidemiological studies have identified small denseLDL as an independent risk factor for CHD [31]which is often associated with both hypertriglyc-eridemia and low HDL-c. However, several reportshave shown that a higher LDL-c/apoB ratio iden-

tifies subjects with predominantly large buoyantLDL particles, whereas a lower value will point topredominantly small dense LDL particles [32—35].Increase in serum apoB may suggest a large num-ber of LDL particles in blood circulation [10]. We

apHtd

S. Biswas et al.

bserved that in patients, increased LDL-c is associ-ted with increased apoB levels. LDL-c to apoB ratioas less than 1.2 in CHD patients found by one group

36], but in our study we observed that in both CHDatients and controls the ratio of LDL-c to apoB isess than 1.2 though the concentration is higher inontrol groups than CHD patients. The apoB/apoAIatio represents the balance of proatherogenic andntiatherogenic lipoproteins and also identifies theipoprotein related risk of vascular disease [37].owever, the ratio of apoB/apoAI was significantlyigher in CHD patients than controls, and it is asso-iated with the disease. According to Sposito et al.DL-c to apoAI ratio was significantly lower in a

iver failure group than controls [38]. In our study,e also found that HDL-c to apoAI ratio is lower inHD patients than controls, but the difference isot statistically significant.

The ROC curve, which is defined as a plot ofest sensitivity versus its 1-specificity was usedo describe and compare the performance ofhe diagnostic test. The AUROC for apoB/apoAI,poB/HDL-c, LDL particle size, TC/HDL-c andonHDL-c/HDL-c are greater than other biochem-cal markers (except apoB) suggesting that theseay provide a better discriminating test for CHD.nd also the difference between the AUROCor apoB and apoB/apoAI (p = 0.832), apoB/HDL-cp = 0.832), LDL particle size (p = 0.516), TC/HDL-(p = 0.118) and nonHDL-c/HDL-c (p = 0.118) were

ot statistically significant. But from these ratiospoB to apoAI ratio is equivalent or better pre-ictive than other lipoprotein and apolipoproteinatios for screening of CHD? The results were eval-ated by ROC curves. Table 3 shows that the ratiof apoB to apoAI and apoB to HDL-c are the bestiscriminators on the basis of ROC curve analysis.hough at a cut-off point sensitivity of apoB/HDL-

showed 11.26% higher rate than apoB/apoAITable 4) and in Table 5, the odds ratio is 16.43or ratio of apoB to HDL-c and 10.41 for ratio ofpoB to apoAI that defines a 57% difference. So,rom the above analysis it is concluded that apoBo apoAI ratio is a good predictive marker for CHD,ut also apoB to HDL-c ratio is a good one andata from Tables 4 and 5 support the apoB toDL-c ratio as clinically superior to apo B/apoAIatio.

However, there are strong correlations betweenach marker and therefore, we performed a mul-ivariate logistic regression analysis and found that

poB, apoB/HDL-c, nonHDL-c, apoB/apoAI and LDLarticle size are contributed in this model. Theosmer—Lemeshow test (p = 0.561) result indicatedhat the number of CHD patients is not significantlyifferent from those predicted by the model, and

L

tnpcostSfdaAaBwttrmtent

tofmarwbatgoroLep

A

TD&stccoBWh

R

[

[

[

[

[

[

[

[17] Krauss RM, Burke DJ. Identification of multiple subclass ofplasma low density lipoproteins in normal humans. J LipidRes 1982;23:97—104.

DL particle size and lipoproteins/apo in CHD

hat the overall model fit is good. The model usingonHDL-c, apoB, apoB/apoAI, apoB/HDL-c and LDLarticle size to predict the probability of being aase or control in the study sample categorized 89f 100 cases (89%) and 86 of 98 controls (87.75%)imilarly. AUROC derived from the model was statis-ically significantly higher from any other variables.o, this model (Fig. 1) is the best discriminatoror CHD. We included patients with acute myocar-ial infarction (AMI), unstable angina and stablengina in our study. We classified the cases ofMI considering WHO and American Heart Associ-tion criteria [39]; unstable angina consistent withraunwald clinical classification [40]. The patientsith stable angina were determined by clinical his-

ory, ECG changes and corroborative evidence ofhe treadmill test, as well as stress echocardiog-aphy. Atherosclerosis and unstable plaque is theajor contributory factor for CAD. We have a limi-

ation in our study regarding the estimation of thextent and burden of atherosclerosis by doing coro-ary angiography and multidetector row computedomography.

Several strengths are present in our study. First,he case—control study has several advantages overther designs, especially, a cohort study. Second,asting bloods were collected enabling valid deter-ination of HDL-c, LDL-c, TG, TC, glucose and

polipoproteins. Third, our study included severalisk factors that have previously not been assessedith conventional risk factors, which might be theest marker of the balance of atherogenic andntiatherogenic particles. In conclusion, since inhe world, there is vast ethnic and cultural hetero-eneity, future investigations should be undertakenn other populations to determine the relativeole of our regression equation (Fig. 1) basedn nonHDL-c, apoB, apoB/apoAI, apoB/HDL-c andDL particle size, which may be useful as mark-rs for diagnosis of predisposition to CHD in aopulation.

cknowledgments

he authors are grateful to Dr Monoj Kar of theepartment of Biochemistry, N.R.S. Medical CollegeHospital, Kolkata, India, for allowing us to use the

emi-auto analyzer and also thankful to the staff ofhe Department of Cardiology, of the same medi-al college for their cooperation during the sampleollection. We are also thankful to the Departmentf Information Centre, Indian Institute of Chemicaliology, Kolkata, for help in analysis of the data.e acknowledge Mr. Ranjit Kumar Das for his kind

elp in this work.

[

125

eferences

[1] Ailhaud G. In: Angel A, Anderson H, Bouchard C, Lau D,Leiter L, Mendelson R, editors. Adipose cell differentia-tion: a long way Tipperary. London: John Libbery & Co. Ltd.;1996. p. 3—11.

[2] Lamarshe B, Després JP, Moorjani S, Cantin B, DegenaisGR, Lupien JR. Prevalence of dyslipidemic phenotypes inischemic heart disease. Am J Cardiol 1995;75:1189—95.

[3] Enas EA. Coronary artery disease epidemic in Indians: acause for alarm and call for action. J Indian Med Assoc2000;98:694—702.

[4] Reddy KS. Cardiovascular disease in India. World Health StatQuart 1993;46:101—7.

[5] Misra A, Vikram NK. Insulin resistance syndrome (metabolicsyndrome) and Asian Indian. Curr Sci 2002;83:1483—96.

[6] Stamler J, Daviglus ML, Garside DB, Dyer AR, Greenland P,Neaton JD. Relationship of baseline serum cholesterol lev-els in 3 large cohorts of younger men to long-term coronary,cardiovascular and all cause mortality and to longevity.JAMA 2000;284:311—8.

[7] Srinivasan SR, Ehnholm C, Wattignecy WA, Bao W, BerensonGS. The relation of apolipoprotein E polymorphism to mul-tiple cardiovascular risks in children: the Bogalusa Heartstudy. Atherosclerosis 1996;123:33—42.

[8] Wiklund O, Angelin B, Olofsson S-O, Ericson M, FagerG, Berglund L, et al. Apolipoprotein (a) and ischemicheart disease in familial hypercholesterolemia. Lancet1990;335:1360—3.

[9] Satoh H, Nishino T, Tomita K, Tsutsui H. Fasting triglycerideis a significant risk factor for coronary artery disease in mid-dle aged Japanese man—–result from 10 year cohort study.Circ J 2006;70:227—31.

10] Larmache B, Lemieux I, Després JP. The small, dense LDLphenotype and the risk of coronary heart disease: epidemi-ology, pathophysiology and therapeutic aspects. DiabetesMetab 1999;25:199—211.

11] Stampfer MJ, Sacks FM, Salvini S, Willent WC, HennekensCH. A prospective study of cholesterol, apolipoproteinsand the risk of myocardial infarction. N Engl J Med1991;35:373—81.

12] Sun YH, Yang YJ, Pei WD, Wu YJ, Gao RL. Patients with lowhigh-density lipoprotein-cholesterol or smoking are morelikely to develop myocardial infarction among subjectswith a visible lesion or stenosis in coronary artery. Circ J2006;70:1602—5.

13] Chan L. Apolipoprotein B, the major protein component oftriglyceride rich and low density lipoproteins. J Biol Chem1992;267:25621—4.

14] Galeano NF, Al-Haideri M, Keyserman F, Rumsey SC, Deckel-baum RJ. Small dense low density lipoprotein has increasedaffinity for LDL receptor independent cell surface bindingsites: a potential mechanism for increased atherogenicity.J Lipid Res 1998;39:1263—73.

15] Kondo A, Muranaka Y, Ohta I, Notsu K, Manabe M, Kotani K,et al. Relationship between triglyceride concentration andLDL size evaluated by malondialdehyde-modified LDL. ClinChem 2001;47:893—900.

16] Varley H, Gowenlock AH, Bell M. Practical clinical biochem-istry. London: William Heinemann Medical Books Ltd.; 1980.pp. 625—684.

18] Matsuzawa Y, Nakamura T, Shimomura I, Kotani K. In: AngelA, Anderson H, Bouchard C, Lau D, Leiter L, Mendelson R,

[

[

[

[

[

[

[

[

[

[

126

editors. Visceral fat accumulation and cardiovascular dis-ease. London: John Liberty & Co., Ltd.; 1996. p. 569—72.

[19] McInnes GT. Hypertension and coronary artery disease:cause and effect. J Hypertens 1995;13:S49—51.

[20] Siekmeier R, Wülfroth P, Wieland H, Gross W, März W.Low density lipoprotein susceptibility to in vitro oxi-dation in healthy smokers and nonsmokers. Clin Chem1996;42:524—30.

[21] Hozawa A, Folsom AR, Sharrett AR, Payne TJ, ChamblessLE. Does the impact of smoking on coronary heart dis-ease differ by low-density lipoprotein cholesterol level? TheAtherosclerosis Risk in Communities (ARIC) Study. Circ J2006;70:1105—10.

[22] Carmena R, Lussier-Cacan S, Roy M, Minnic A, LingenhelA, Kronenberg F, et al. Lp(a) levels and atheroscleroticvascular disease in a sample of patients with familial hyper-cholesterolemia sharing the same gene defect. ArteriosclerThromb Vasc Biol 1996;16:129—36.

[23] Lewis GF, Steiner G. Hypertriglyceridemia and its metabolicconsequences as a risk factor for atherosclerotic cardiovas-cular disease in non-insulin-dependent diabetes mellitus.Diabetes 1996;12:36—56.

[24] Masunaga N, Kimura A, Miyatak M, Nishioka N, Hirano Y,Hayashi T, et al. Effects of alcohol consumption on cardio-vascular events in male patients with healed myocardialinfarction. Circ J 2006;70:1263—8.

[25] Gandhi BM. Lipoprotein composition of normal healthysubjects in Northern India. Indian J Med Res 1982;75:393—401.

[26] Natio HK. Coronary artery disease & disorders of lipidmetabolism. In: Kaplan A, Pesce AJ, editors. Clinical chem-istry. St. Louis: Mosby Publishing; 1996. p. 642—82.

[27] Stain EA, Myers GL. Lipids, apolipoproteins and lipopro-teins. In: Burtis CA, Ashwood ER, editors. Tietz funda-mentals of clinical chemistry. Philadelphia: WB SaundersCompany; 1996. p. 375—401.

[28] Sun YH, Yang YJ, Pie WD, Wu YJ, Gao RL. Patients withlow high-density lipoprotein cholesterol or smoking aremore likely to develop myocardial infarction among sub-jects with a visible lesion or stenosis in coronary artery.

Circ J 2006;70:1602—5.

[29] Panagiotakos DB, Pitsavos C, Skoumas J, Chrysohoou C,Toutouza M, Stefanadis CI, et al. Importance of LDL/HDLcholesterol ratio as a predictor for coronary heart diseaseevents in patients with heterozygous familial hypercholes-

[

Available online at www.s

S. Biswas et al.

terolaemia: a 15-year follow-up (1987—2002). Curr Med ResOpin 2003;19:89—94.

30] Maruyama C, Imamura K, Teramoto T. Assessment of LDLParticle Size by Triglyceride/HDL Cholesterol Ratio inNon-diabetic Healthy Subjects without Prominent Hyper-lipidemia. J Atheroscler Thromb 2003;10:186—91.

31] Krauss RM. Heterogenicity of plasma low-densitylipoproteins and atherosclerosis risk. Curr Opin Lipidol1994;5:339—49.

32] Wagner AM, Jorba D, Rigla M, Alonso E, Ordonez-Llanos J,Perez A. LDL cholesterol/apolipoprotein B ratio is a goodpredictor of LDL phenotype B in type 2 diabetes. Acta Dia-betol 2002;39:215—20.

33] Sniderman AD, Lamarche B, Tilley J, Secombe DW, ForhlichJ. Hypertriglyceridemia hyperapo B in type 2 diabetes. Dia-betes Care 2002;25:579—82.

34] Sniderman AD, Dagenasis GR, Cantin B, Despres JP, LamarchB. High lipoprotein B with low high density lipoproteincholesterol and normal plasma triglycerides and choles-terol. Am J Cardiol 2001;87:792—3.

35] Sniderman AD, St. Pierre A, Cantin B, Dagenais GR, DeprésJP, Lamarche B. Concordance/discordance between plasmaapolipoprotein B levels and the cholesterol indexes ofatherosclerotic risk. Am J Cardiol 2003;91:1173—7.

36] Enas EA, Senthilkumar A. Coronary artery disease in AsianIndians: an update and review. Internet J Cardiol 2001;1(2).

37] Sniderman AD, Jungner I, Holme I, Aastveit A, Walldius G.Errors that result from using the TC/HDL-c ratio rather thanthe apoB/apoA-I ratio to identify the lipoprotein relatedrisk of vascular disease. J Intern Med 2006;259:455—61.

38] Sposito AC, Vinagre CG, Pandullo FL, Mies S, Raia S, RaniresJAF. Apolipoprotein and lipid abnormalities in chronic liverfailure. Braz J Med Biol Res 1997;30:1287—90.

39] Luepker RV, Apple FS, Christenson RH, Crow RS, Fort-mann SP, Goff D, et al. Case definitions for acute coronaryheart disease in epidemiology and clinical research stud-ies: a statement from the AHA Council on Epidemiology andPrevention; AHA Statistics Committee; World Heart Federa-tion Council on Epidemiology and Prevention; the EuropeanSociety of Cardiology Working Group on Epidemiology and

Prevention; Centers for Disease Control and Prevention; andthe National Heart, Lung, and Blood Institute. Circulation2003;108:2543—9.

40] Hamm CW, Braunwald E. A classification of unstable anginarevisited. Circulation 2000;102:118—22.

ciencedirect.com