- Email: [email protected]
LDL-apo B as indices of hyperapobetalipoproteinemia and small dense LDL
Atherosclerosis 138 (1998) 289 – 299
Development of approximate formula for LDL-chol, LDL-apo B and LDL-chol/LDL-apo B as indices of hyperapobetalipoproteinemia and small dense LDL Yuichi Hattori c, Masaaki Suzuki a, Motoo Tsushima a, Masami Yoshida a, Yoko Tokunaga a, Ying Wang a, Di Zhao a, Makoto Takeuchi d, Yasushi Hara a, Kayoko Ikeda Ryomoto a, Motoyoshi Ikebuchi a, Hiroshi Kishioka e, Toshifumi Mannami b, Syunnroku Baba b, Yutaka Harano a,* a
Di6ision of Atherosclerosis, Metabolism and Clinical Nutrition, National Cardio6ascular Center (NCVC), 5 -7 -1 Fujishiro-dai, Suita, Osaka 565, Japan b Di6ision of Pre6enti6e Medicine, National Cardio6ascular Center, 5 -7 -1 Fujishiro-dai, Suita, Osaka 565, Japan c Konan Uni6ersity, Kobe 658, Japan d Kobe City College of Technology, Kobe 651 -21, Japan e Kyowa Medix KK, Tokyo, Japan Received 14 February 1997; received in revised form 5 January 1998; accepted 23 January 1998
Abstract Estimation of LDL-chol and LDL-apo B is useful for the diagnosis of hyperapobetalipoproteinemia (normal LDL-chol with increased LDL-apo B), which is one of the most commonly occurring lipoprotein disorders associated with atherosclerotic cardiovascular diseases. The LDL-chol/LDL-apo B ratio reflects the level of small dense LDL, which is an important risk factor for IHD, CVD and ASO. In order to estimate LDL-apo B and LDL-chol/LDL-apo B ratio from blood chol, TG, HDL-chol and apo B values, we developed a formula for LDL-chol {0.94Chol −0.94HDL-chol− 0.19TG}, LDL-apo B {apo B − 0.09Chol+ 0.09HDL-chol − 0.08TG}, and LDL-chol/LDL-apo B [{0.94Chol− 0.94HDL-chol−0.19TG}/{apo B − 0.09Chol+0.09HDLchol− 0.08TG}] using ultracentrifugal data from 2179 subjects. These were calculated by the least squares method on the assumption that a certain compositional relationship exists between Chol, TG and apo B in VLDL, IDL and LDL. Friedewald’s formula for LDL-chol (Chol− HDL-chol− 0.2TG) includes IDL-chol, but the present new formula theoretically excludes IDL-chol. It suggests a better estimation for the correct LDL-chol. Estimated LDL-apo B is useful for the diagnosis of hyperapobetalipoproteinemia and detection of small dense LDL. Without performing ultracentrifuge, additional information is obtained for the quantitative and qualitative alteration of LDL, such as small dense LDL. The above formulae and a new classification of lipoproteinemia including apo B were applied to the analyses of lipoprotein profiles of subjects with cardiovascular diseases, which were compared with those in the general population. Hyperapobetalipoproteinemia with high TG was observed 2–3 times more frequently in subjects with CAD, MI and ASO than in the Suita population. Lower ratios of LDL-chol/LDL-apo B, reflecting preponderance of small dense LDL, were observed in the above three groups. Type IIb and combined low HDL-chol were also frequent phenotypes in CAD, A-Th and ASO. The present formulae are useful for the detailed analyses of lipoprotein disorders in both qualitative as well as quantitative aspects. © 1998 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Hyperapobetalipoproteinemia; LDL-apo B; LDL-chol; Small dense LDL
* Corresponding author. Tel: +81 6 8335012; fax: + 81 6 8728100. 0021-9150/98/$19.00 © 1998 Elsevier Science Ireland Ltd. All rights reserved. PII S0021-9150(98)00034-3
290
Y. Hattori et al. / Atherosclerosis 138 (1998) 289–299
Fig. 1. (a) Correlation of LDL-chol values between measured and estimated method. LDL-chol was calculated according to: LDLC =0.94Chol − 0.94HDLC −0.19TG; (b) Correlation of LDL-chol values between measured and estimated method using Friedewald’s equation. LDL-chol was calculated according to Friedewald’s equation: LDLC = Chol−HDLC −0.2TG.
Y. Hattori et al. / Atherosclerosis 138 (1998) 289–299
291
Fig. 2. Correlation of LDL-apo B values between measured and estimated method. LDL-apo B was calculated according to: LDLB =apo B − 0.09Chol+ 0.09HDLC −0.08TG.
1. Introduction In addition to the elevated levels of LDL-chol and TG, and low HDL-chol [1], increased apo B content in plasma gives valuable information regarding atherogenic lipoprotein profiles [2]. Sniderman et al. first reported hyperapobetalipoproteinemia which is characterized by increased LDL-apo B with normal LDL-chol often observed in myocardial infarction [3]. Small dense LDL, which has a property of low LDL-chol/LDL-apo B ratio, and is known to be also characteristic of hyperapobetalipoproteinemia, has been reported as less cleared from the blood stream and more susceptible for conversion into atheromatous lesion. Dyslipidemia has been reported in diabetes [4,5], obesity [6], combined hyperlipidemia [7] and in cardiovascular diseases [8]. These abnormalities have been noted even when plasma levels of Chol and TG are normal. Kwiterovich et al. reported that hyperapobetalipoproteinemia was the most common phenotype (34%) of hyperlipidemia in subjects with premature coronary artery disease (CAD) [9]. We have developed approximate formulae for LDLchol and LDL-apo B, which provide the frequency of
hyperapobetalipoproteinemia and small dense LDL (low LDL-chol/LDL-apo B ratio) using plasma levels of Chol, TG, HDL-chol and apo B. Practical application has been attempted in subjects with cardiovascular diseases who had been admitted and the calculated data were compared with those obtained in a population-based field study.
2. Materials and methods
2.1. Subjects Blood was obtained from 2161 out- or inpatients with hyperlipidemia, obesity, diabetes, hypertension or atherosclerotic diseases and 18 healthy controls during the past 8 years. Those with chol and TG over 350 and 400, less than 100 and 30, and apo B less than 40 mg/dl were all excluded. Among these, 18 healthy control subjects were included who have none of the above diseases. Those with liver and renal dysfunction were also excluded (Table 1). Ischemic heart disease consists of
Y. Hattori et al. / Atherosclerosis 138 (1998) 289–299
292
Fig. 3. Correlation of LDL-chol/LDL-apo B values between measured and estimated method. The ratio LDL-chol/LDL-apo B was calculated according to: LDLC LDLB
=
0.94Chol − 0.94HDLC − 0.19TG . apo B −0.09Chol + 0.09HDLC − 0.08TG
myocardial infarction (MI), and coronary artery disease (CAD). MI was diagnosed if elevation of enzymes, typical ECG and UCG changes were present. CAD was diagnosed when more than 75% stenosis was noted in coronary arteries on catheterization after screening with positive stress tests. Cerebral atherothrombotic infarction was diagnosed on clinical symptoms, CT, MRI and angiography. Diagnosis of ASO was based on claudication, ankle pressure index below 0.9, and angiography. These atherosclerotic subjects were admitted to the Atherosclerosis, Metabolism and Clinical Nutrition wards between 1992 and 1995. Subjects for the population-based Suita study were recruited from people who had been receiving regular health check-ups between April 1995 and February 1996, in The Department of Preventive Cardiology, NCVC. Non-hospitalized subjects participated on a voluntary basis. Their ages ranged between 30 and 79 years old, and they were postulated to be adequate as control subjects.
2.2. Blood samples and ultracentrifugal procedures Blood samples were obtained in EDTA-containing tubes from the subjects following an overnight fast. Plasma was immediately separated and kept at 4°C. VLDL, IDL, LDL and HDL subfractions were separated by ultracentrifugation using a Beckman TL-100 essentially according to the method described by Hatch Table 1 Patient profiles (mean 9S.E.)
Age (years) Gender (M/F) Chol (mg/dl) TG (mg/dl) HDLC (mg/dl) Apo B (mg/dl)
Healthy controls (n =18)
Patients (n = 2161)
49.2 95.7 7/11 186.2 97.2 77.0 95.9 61.3 93.3 86.0 94.0
57.9 9 0.27 1151/1010 207.5 90.93 130.0 91.4 44.4 90.32 121.3 90.73
Y. Hattori et al. / Atherosclerosis 138 (1998) 289–299
293
Fig. 4. (a) Correlation of LDL-chol values between measured and estimated method for healthy control subjects; (b) Correlation of LDL-apo B values between measured and estimated method for healthy control subjects; (c) Correlation of LDL-chol/LDL-apo B values between measured and estimated for healthy control subjects.
294
Y. Hattori et al. / Atherosclerosis 138 (1998) 289–299
Fig. 4. (Continued)
and Lee [10]. Those with obvious chylomicronemia (floating cake layer) were excluded. Chol and TG were determined by enzymatic method (Kyowa Medix, Tokyo) [11] as described before.
2.3. Apo B determination by the latex method A 10% aqueous suspension of a polystyrene latex (particle size 0.2 mm) was diluted 10-fold with phosphate buffer (pH 7.2). To 10 ml of the aqueous suspension of polystyrene latex, 2 ml of 1 mg/ml solution of the anti-apolipoprotein B specific purified antibody and 0.5 ml of 1 mg/ml solution of 1-ethyl-3 (3-dimethylaminopropyl)-carbodiimide-hydrochloride were added and the mixture reacted for 4 h at 5°C. The mixture was centrifuged at 10000 rev./min for 1 h to remove the excess anti-apolipoprotein B specific purified antibody. The antibody-coated polystyrene latex thus obtained was suspended in 50 ml of a glycine buffer (pH 8.2) containing 0.3% bovine serum albumin. To this aliquot (250 ml), 220 ml of 0.1 M glycine buffer (pH 8.6) was added, then 30 ml specimen and the mixture stirred. The turbidity and the transmission at 3.5 min was determined with an automated immunoassay system (EL-1000, Kyowa Medix, Japan). The method is partic-
ularly sensitive and the plasma and LDL were diluted 100-fold and the other subfractions were diluted 30-fold before the assay. CV was between 1.5–2.5%. Since apo B antibody which was raised against human LDL was polychronal, standard curves for VLDL, IDL, LDL, were identical [12]. Although chylomicron behaved in a way similar to LDL, obvious chylomicronemia was excluded from the study. Therefore, both apo B-100 and apo B-48 were detectable by the present method. The latex method is more accurate compared with immunoturbidity method which tends to produce a higher value for samples with high TG [4,6,11,12]. Apo B values determined by the latex method correlate closely with those produced by enzyme immunoassay [13]. In contrast to the commercially available immuno-turbidity method, which shows erroneously higher VLDL-apo B value in hypertriglyceridemia, the latex method gives accurate values for apo B in both VLDL and LDL.
2.4. Mathematical procedures From the thousands of data in lipoprotein subfractions, we noted that a constant relation exists among TG, Chol and apo B content in VLDL, IDL and LDL.
1514 43 46 20 42
674/840 30/13 32/14 13/7 33/9
Gender
22.790.1 24.4 9 0.5** 24.99 0.5** 22.0 90.7 22.4 9 0.4
BMI
Values are mean 9S.E. (mg/dl). By t-test vs control, *PB0.05; **PB0.01.
Control IHD CAD MI A-Th ASO
No of subjects 21090.8 2299 7.0** 2079 6.5 21998.9 21896.1
Chol
HDLC 59.1 9 0.4 47.9 92.1** 43.8 91.9** 51.89 3.6** 49.29 2.3**
TG
112 92.5 172 913.2** 146 99.0* 154 9 20.0 141 910.9*
Table 2 Lipoprotein profiles for control and macroangiopathy (MA: IHD, CVD, ASO)
98.0 90.9 134 9 4.7** 126 9 5.0** 117 9 7.6* 124 9 4.4**
Apo B
17.6 9 0.3 25.6 92.1** 21.6 9 1.4* 22.9 9 2.4* 21.8 9 1.7*
VLDLC
0.149 0.0003 0.14 9 0.003 0.15 90.001 0.15 9 0.009 0.15 90.001
VLDLC/TG
119.4 9 0.8 136.2 9 6.2** 124.0 95.6 126.5 9 7.5* 127.3 95.1**
LDLC
91.89 0.7 111.49 4.1* 105.594.4 96.89 6.4 104.093.7*
LDLB
1.39 90.01 1.26 90.06* 1.219 0.04** 1.36 90.07 1.25 90.05*
LDLC/LDLB
Y. Hattori et al. / Atherosclerosis 138 (1998) 289–299 295
Y. Hattori et al. / Atherosclerosis 138 (1998) 289–299
296
Fig. 5. New classification of hyperlipoproteinemia including apo B. Table 3 Association between lipoprotein phenotype and macroangiopathy Lipoprotein phenotypes
Hyperapo B (all) Hypertriglyceridemic Normotriglyceridemic Type IIa Type IIb Type IV Normal Hypoalphaproteinemia Total Combined low-HDLC
Control
275 89 186 566 187 101 371 14
(18.2) (5.9) (12.3) (37.4) (12.3) (6.7) (24.5) (0.9)
1514 107 (7.1)
IHD CAD
MI
10 7 8 10 19 1 3 0
16 9 7 10 8 4 7 1
(23.3) (16.3)* (18.6) (23.3) (44.2)** (2.3)** (7.0)** (0)**
43 11 (25.6)**
A-Th (34.8)** (19.6)* (15.2) (21.7)* (17.4) (8.7)** (15.2)** (2.2)**
46 21 (45.7)**
3 1 2 3 7 2 4 1
(15.0) (5.0) (10.0) (15.0) (35.0)** (10.0)** (20.0)** (5.0)**
20 5 (25.0)**
ASO 9 6 3 12 12 1 8 0
(21.4) (14.3)* (7.1)** (28.6) (28.6)** (2.4)** (19.0)** (0)**
42 11 (26.2)**
Values are number of subjects with percentage in parentheses. By Chi square analysis, *PB0.05, **PB0.01.
HDL-apo B includes LP(a) and cannot be ignored. Theoretically the following equations are given for total plasma cholesterol and apo B concentration: Chol = VLDLC + IDLC +LDLC +HDLC; apo B= VLDLB + IDLB +LDLB +HDLB; where subscript C is Chol, and subscript B is apo B. The following formulae are assumed with constant a and b for Chol and TG: VLDLC +IDLC = a(Chol −HDLC) +bTG; VLDLB +IDLB + HDLB =a(Chol −HDLC) +bTG. Regression analyses using the actual data have been carried out using SAS [14,15] (software package for statistical analyses). LDLC and LDLB were obtained by
subtracting the above values from plasma Chol other than HDLC and apo B.
3. Results
3.1. Estimation of LDL-chol Regression analyses using the equation shown in Section 2.4 have been carried out with SAS and have produced the following formula: VLDLC + IDLC = 0.06(Chol− HDLC)+ 0.19TG
(1)
Y. Hattori et al. / Atherosclerosis 138 (1998) 289–299
The correlation between estimated (VLDLC +IDLC) and measured values is 0.71. The following formula was then obtained after subtracting the above value and HDLC from total chol:
297
0.94Chol− 0.94HDLC − 0.19TG LDLC = LDLB apo B− 0.09Chol+ 0.09HDLC − 0.08TG A significant correlation was noted between measured and estimated values (r= 0.67, PB 0.001) (Fig. 3).
LDLC = Chol− HDLC −VLDLC −IDLC = 0.94Chol−0.94HDLC −0.19TG
(2)
The value LDLC was calculated for each subject according to Eq. (2). Plots of each LDLC calculated by this method versus that obtained with actual ultracentrifugation are shown (Fig. 1A). A good correlation (r= 0.86) between estimated and measured LDLC was observed. A somewhat better, but similar correlation was noted by the present equation compared with Friedewald’s [16], although the present value for LDLC does not theoretically include IDLC (Fig. 1B).
3.2. Estimation of VLDL-chol The following formula is assumed for VLDLC, with constant a, b for Chol and TG. VLDLC =a(Chol− HDLC) +bTG Regression analyses on this model have been carried out using the SAS program, which derive the following formula: VLDLC = − 0.01(Chol−HDLC) +0.16TG
(3)
The value VLDLC was calculated for each subject according to Eq. (3). A significant correlation between estimated VLDLC and the measured value was noted with r = 0.81 (PB0.0001, data not shown). VLDL-chol is valuable for the diagnosis of type III hyperlipidemia (VLDL-chol/TG \ 0.3).
3.3. Estimation of LDL-apo B Using individual data, regression analyses by SAS gave the following equation: VLDLB +IDLB + HDLB = 0.09(Chol−HDLC) +0.08TG A significant correlation between estimated (VLDLB +IDLB + HDLB) and measured value was noted with r =0.61 (P B0.001, Fig. not shown). Finally, LDLB was obtained as follows:
3.4. Calculated lipoprotein analyses in control healthy subjects LDL-chol in control healthy subjects falls between 50 and 130 excluding two cases. A high correlation (r= 0.95) was observed between measured and estimated values (Fig. 4A). Calculated LDL-apo B ranges from 40 to 90 mg/dl (90% percentile) with good correlation with the actually measured values (Fig. 4B). The ratio of the above estimated LDL-chol/LDL-apo B in the control subjects is shown in Fig. 4C. Normal values range from 1.2 to 2.3 (90% percentile). This ratio reflects the size of LDL and the preponderance of small dense particles is postulated when the ratio falls below 1.2.
3.5. Calculated lipoprotein analyses in subjects with atherosclerotic cardio6ascular diseases and population-based Suita study In CAD, in addition to the elevation of serum Chol, TG and apo B with low HDL-chol, increased LDL-apo B and LDL-chol and lower ratio of LDL-chol/LDL-apo B were noted, indicating the increased number of smaller size of LDL (Table 2). In MI the same abnormality was noted except that the elevation of LDL-apo B was not statistically significant when compared with normal LDL. In ASO hyperapobetalipoproteinemia was observed, since normal LDL-chol and increased LDL-apo B and lowered LDL-chol/LDL-apo B ratio were noted. In atherothrombotic cerebral infarction, elevation of serum apo B with low HDL-chol and increased VLDLchol and LDL-chol were noted. Using the above lipoprotein profiles and new classification including apo B (Fig. 5), the above subjects with atherosclerosis were classified into the following phenotypes (Table 3). Hyperapobetalipoproteinemia with high TG was three times more frequent in CAD, MI and ASO over Suita population. Type IIb was 2.5–3.5 times more frequent in CAD, A-Th and ASO while the other phenotypes were less remarkable and normolipidemia was significantly less in CAD, A-Th and ASO. Low HDL-chol with combined other phenotypes was 3- to 7-fold more prevalent in cardiovascular diseases.
LDLB = apo B− VLDLB −IDLB −HDLB = apo B− 0.09Chol +0.09HDLC −0.08TG The close correlation (r = 0.87) between estimated and actually measured LDLB values is shown in Fig. 2. LDL-chol/LDL-apo B can be calculated from the above formulae as follows:
4. Discussion We developed estimated formulae for LDL-chol, LDL-apo B, VLDL-chol and LDL-chol/LDL-apo B using plasma Chol, TG, HDL-chol and apo B. Ultracen-
298
Y. Hattori et al. / Atherosclerosis 138 (1998) 289–299
trifugal data from 2179 subjects were used for obtaining the above formula. The equation has been developed under the assumption that VLDLC +IDLC is the function of (Chol−HDLC) and TG. Then LDL was calculated by subtracting VLDLC +IDLC and HDLC from total chol. The value thus obtained theoretically does not contain IDL while Friedewald’s formula for LDLchol (Chol−HDL-chol− 0.2TG) contains IDL-chol. Actually, Friedewald’s formula gives somewhat higher results than the present values (Fig. 1A,B). LDL-apo B has been calculated based on the assumption that VLDLB +IDLB + HDLB is the function of (Chol − HDL-chol) and TG and the correlation with the actual value was highly significant (r =0.87, P B 0.001). LDLapo B value is necessary for the diagnosis of hyperapobetalipoproteinemia (HBL), which is the most common lipoprotein disorder in cardiovascular diseases [17]. A small dense LDL, which shows a low ratio of LDLC/ LDLB is also a disorder characteristic of HBL. This disorder is often observed in diabetes [4], obesity [5], HT [18], and hyperlipidemia as well as in cardiovascular diseases [8,19]. The difference between the estimated and measured values divided by the actual mean has been plotted against measured values. The percent samples which fall between −0.3 0.3 is 66.7% for LDLchol, 84.3% for LDL-apo B and 86.0% for LDL-chol/LDL-apo B, respectively. The above values for LDL-chol (66.7%) and the others were higher than the Friedewald’s formula (51.0%). These data suggest that although the estimated formula has some limitations, it can be used for the analysis of lipoprotein disorder. Tests of the significance of b for the regression equations were all at the 1% level. A new classification of hyperlipidemia using apo B has been advocated by Kwiterovich et al. [9]. Using the present formula for LDLC and LDLB, this new classification can be applied. Without performing ultracentrifugal separation, additional information is obtained for the quantitative and qualitative alteration of LDL, which is considered to be a risk for cardiovascular diseases. This classification is more suitable than the classification of WHO since HBL was noted in 20 – 40% of patients with cardiovascular disease. The strict definition of HBL is normal LDL-chol with elevated LDL-apo B, which is often observed in myocardial infarction [8] as well as in acute coronary syndrome and characteristic disorders shown in obesity, diabetes and hypertension [18]. In fact, HBD with high TG was 2.5 – 3 times more frequent in CAD, MI and ASO. The characteristic feature of small dense LDL, lower ratios of LDL-chol/LDL-apo B were observed in these three groups. The present formulae are valuable for the detailed analysis of lipoprotein disorder since additional information regarding qualitative as well as quantitative properties of lipoprotein can be obtained, although there is the limitation that this is an estimation, not the actual value.
Acknowledgements This study was supported in part by Special Coordination Funds for promoting Science and Technology (Encouragement System of COE) and from a grant for Scientific Research Expenses for Health and Welfare Programs (Clinical Treatment of Diabetes Mellitus, Akanuma).
References [1] Castelli WP, Garrison RJ, Wilson PWP, Abbott RD, Kalousdian S, Kannel WB. Incidence of coronary heart disease and lipoprotein cholesterol levels. The Framingham Heart Study 1986. J Am Med Assoc 1986;256:2835. [2] Zhao X-Q, Brown BG, Hillger L, Sacco D, Bisson B, Fisher L, Albers JJ. Effects of intensive lipid-lowering therapy on the coronary arteries of asymptomatic subjects with elevated apolipoprotein B. Circulation 1993;88:2744. [3] Sniderman AD, Wolfson C, Teng B, Franklin FA, Bachorik PS, Kwiterovich PO Jr. Association of hyperapobetalipoproteinemia with endogenous hypertriglyceridemia and atherosclerosis. Ann Intern Med 1982;97:833. [4] Harano Y, Kageyama A, Nakao Y, Hara Y, Suzuki M, Sato A, Ikebuchi M, Shinozaki K, Tsushima M. Quantitative and qualitative derangement of apolipoprotein B-containing lipoproteins as a risk factor for diabetic macroangiopathy in nonobese NIDDM subjects. Diabetes 1996;45(Suppl):S31. [5] Lanhdenpera¨ S, Syva¨nne M, Kahri J, Taskinen M-R. Regulation of low-density lipoprotein particle size distribution in NIDDM and coronary disease — importance of serum triglycerides. Diabetologia 1996;39:453. [6] Harano Y, Nakao Y, Kageyama A, Suzuki M, Hirose J, Asakura Y, Sato A, Komatsu R, Tsusima M, Yamamoto A. A contribution of hyperinsulinemia and hyperglycemia to lipoprotein disorder in obesity. In: Oomura Y, Tarui S, Inoue S, Shimazu T, editors. Progress in Obesity Research. London: Libby, 1990:299 – 302. [7] Schwandt P. The triglyceride controversy — a review of the data. Eur Heart J 1990;11(Suppl H):43. [8] Yokota C, Nonogi H, Miyazaki S, Goto Y, Haze Y, Hara Y, Shinozaki K, Ikebuchi M, Suzuki M, Harano Y. Lipoprotein analyses in patients with stable angina and acute coronary syndrome. Int J Cardiol 1996;57:161. [9] Kwiterovich PO Jr, Coresh J, Bachorik PS. Prevalence of hyperapobetalipoproteinemia and other lipoprotein phenotypes in men (aged 5 50 years) and women (aged 560 years) with coronary artery disease. Am J Cardiol 1993;71:631. [10] Hatch FT, Lee RS. Practical method for plasma lipoprotein analysis. Adv Lipid Res 1968;6:1. [11] Ryomoto KI, Suzuki M, Tsushima M, Yamamoto A, Harano Y. Effects of simvastatin on the number and composition of apoprotein B-containing lipoproteins in hypercholesterolemia: analysis of apoprotein B in each lipoprotein fraction by highly sensitive latex method. Endocr J 1996;43:469. [12] Harano Y. Highly sensitive apo B assay and its clinical significance. Jpn J Clin Pathol 1990;38:861. [13] Harano Y, Kojima H, Nakano T, Harada M, Kashiwagi A, Nakajima T, Hidaka TH, Ohtsuki T, Suzuki T, Tamura A, Fujii T, Nishimura T, Ohtaka T, Shigeta Y. Homozygous hypobetalipoproteinemia with spared chylomicron formation. Metabolism 1989;38:1.
Y. Hattori et al. / Atherosclerosis 138 (1998) 289–299 [14] Freund J, Little RC. SAS System for Regression. 2nd ed. Cary, NC: SAS Institute, 1992. [15] Schlotzhauer SD, Little RC. SAS System for Elementary Statistical Analysis. Cary, NC: SAS Institute, 1991. [16] Friedewald T, Levy RI, Fredrickson DS. Estimation of the concentration of low density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem 1972;18:499. [17] Kwiterovich PO Jr, Motevalli M, Woodford FP, Davignon J, Sniderman A, editors. Hyperapo B—An Update. Atherosclero-
.
299
sis X. Amsterdam: Elsevier, 1995:966 – 971. [18] Ikeda K, Suzuki M, Ikebuchi M, Hara Y, Tsushima M, Yamamoto A, Harano Y. Hyperbetalipoproteinemia with small low-density lipoprotein, a characteristic disorder of lipoprotein in essential hypertension. J Diabetes Complications 1995;9:227. [19] Sato A, Kanazawa A, Hasegawa M, Asakura Y, Yokota C, Suzuki M, Tusima M, Kuriyama Y, Sawada T, Yamaguchi T, Kobayashi M, Harano Y. Lipoprotein disorder in brain infarction and hemorrhage. Int J Clin Lab Res 1998 (in press).
.
Development of approximate formula for LDL-chol, LDL-apo B and LDL-chol/LDL-apo B as indices of hyperapobetalipoproteinemia and small dense LDL Yuichi Hattori c, Masaaki Suzuki a, Motoo Tsushima a, Masami Yoshida a, Yoko Tokunaga a, Ying Wang a, Di Zhao a, Makoto Takeuchi d, Yasushi Hara a, Kayoko Ikeda Ryomoto a, Motoyoshi Ikebuchi a, Hiroshi Kishioka e, Toshifumi Mannami b, Syunnroku Baba b, Yutaka Harano a,* a
Di6ision of Atherosclerosis, Metabolism and Clinical Nutrition, National Cardio6ascular Center (NCVC), 5 -7 -1 Fujishiro-dai, Suita, Osaka 565, Japan b Di6ision of Pre6enti6e Medicine, National Cardio6ascular Center, 5 -7 -1 Fujishiro-dai, Suita, Osaka 565, Japan c Konan Uni6ersity, Kobe 658, Japan d Kobe City College of Technology, Kobe 651 -21, Japan e Kyowa Medix KK, Tokyo, Japan Received 14 February 1997; received in revised form 5 January 1998; accepted 23 January 1998
Abstract Estimation of LDL-chol and LDL-apo B is useful for the diagnosis of hyperapobetalipoproteinemia (normal LDL-chol with increased LDL-apo B), which is one of the most commonly occurring lipoprotein disorders associated with atherosclerotic cardiovascular diseases. The LDL-chol/LDL-apo B ratio reflects the level of small dense LDL, which is an important risk factor for IHD, CVD and ASO. In order to estimate LDL-apo B and LDL-chol/LDL-apo B ratio from blood chol, TG, HDL-chol and apo B values, we developed a formula for LDL-chol {0.94Chol −0.94HDL-chol− 0.19TG}, LDL-apo B {apo B − 0.09Chol+ 0.09HDL-chol − 0.08TG}, and LDL-chol/LDL-apo B [{0.94Chol− 0.94HDL-chol−0.19TG}/{apo B − 0.09Chol+0.09HDLchol− 0.08TG}] using ultracentrifugal data from 2179 subjects. These were calculated by the least squares method on the assumption that a certain compositional relationship exists between Chol, TG and apo B in VLDL, IDL and LDL. Friedewald’s formula for LDL-chol (Chol− HDL-chol− 0.2TG) includes IDL-chol, but the present new formula theoretically excludes IDL-chol. It suggests a better estimation for the correct LDL-chol. Estimated LDL-apo B is useful for the diagnosis of hyperapobetalipoproteinemia and detection of small dense LDL. Without performing ultracentrifuge, additional information is obtained for the quantitative and qualitative alteration of LDL, such as small dense LDL. The above formulae and a new classification of lipoproteinemia including apo B were applied to the analyses of lipoprotein profiles of subjects with cardiovascular diseases, which were compared with those in the general population. Hyperapobetalipoproteinemia with high TG was observed 2–3 times more frequently in subjects with CAD, MI and ASO than in the Suita population. Lower ratios of LDL-chol/LDL-apo B, reflecting preponderance of small dense LDL, were observed in the above three groups. Type IIb and combined low HDL-chol were also frequent phenotypes in CAD, A-Th and ASO. The present formulae are useful for the detailed analyses of lipoprotein disorders in both qualitative as well as quantitative aspects. © 1998 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Hyperapobetalipoproteinemia; LDL-apo B; LDL-chol; Small dense LDL
* Corresponding author. Tel: +81 6 8335012; fax: + 81 6 8728100. 0021-9150/98/$19.00 © 1998 Elsevier Science Ireland Ltd. All rights reserved. PII S0021-9150(98)00034-3
290
Y. Hattori et al. / Atherosclerosis 138 (1998) 289–299
Fig. 1. (a) Correlation of LDL-chol values between measured and estimated method. LDL-chol was calculated according to: LDLC =0.94Chol − 0.94HDLC −0.19TG; (b) Correlation of LDL-chol values between measured and estimated method using Friedewald’s equation. LDL-chol was calculated according to Friedewald’s equation: LDLC = Chol−HDLC −0.2TG.
Y. Hattori et al. / Atherosclerosis 138 (1998) 289–299
291
Fig. 2. Correlation of LDL-apo B values between measured and estimated method. LDL-apo B was calculated according to: LDLB =apo B − 0.09Chol+ 0.09HDLC −0.08TG.
1. Introduction In addition to the elevated levels of LDL-chol and TG, and low HDL-chol [1], increased apo B content in plasma gives valuable information regarding atherogenic lipoprotein profiles [2]. Sniderman et al. first reported hyperapobetalipoproteinemia which is characterized by increased LDL-apo B with normal LDL-chol often observed in myocardial infarction [3]. Small dense LDL, which has a property of low LDL-chol/LDL-apo B ratio, and is known to be also characteristic of hyperapobetalipoproteinemia, has been reported as less cleared from the blood stream and more susceptible for conversion into atheromatous lesion. Dyslipidemia has been reported in diabetes [4,5], obesity [6], combined hyperlipidemia [7] and in cardiovascular diseases [8]. These abnormalities have been noted even when plasma levels of Chol and TG are normal. Kwiterovich et al. reported that hyperapobetalipoproteinemia was the most common phenotype (34%) of hyperlipidemia in subjects with premature coronary artery disease (CAD) [9]. We have developed approximate formulae for LDLchol and LDL-apo B, which provide the frequency of
hyperapobetalipoproteinemia and small dense LDL (low LDL-chol/LDL-apo B ratio) using plasma levels of Chol, TG, HDL-chol and apo B. Practical application has been attempted in subjects with cardiovascular diseases who had been admitted and the calculated data were compared with those obtained in a population-based field study.
2. Materials and methods
2.1. Subjects Blood was obtained from 2161 out- or inpatients with hyperlipidemia, obesity, diabetes, hypertension or atherosclerotic diseases and 18 healthy controls during the past 8 years. Those with chol and TG over 350 and 400, less than 100 and 30, and apo B less than 40 mg/dl were all excluded. Among these, 18 healthy control subjects were included who have none of the above diseases. Those with liver and renal dysfunction were also excluded (Table 1). Ischemic heart disease consists of
Y. Hattori et al. / Atherosclerosis 138 (1998) 289–299
292
Fig. 3. Correlation of LDL-chol/LDL-apo B values between measured and estimated method. The ratio LDL-chol/LDL-apo B was calculated according to: LDLC LDLB
=
0.94Chol − 0.94HDLC − 0.19TG . apo B −0.09Chol + 0.09HDLC − 0.08TG
myocardial infarction (MI), and coronary artery disease (CAD). MI was diagnosed if elevation of enzymes, typical ECG and UCG changes were present. CAD was diagnosed when more than 75% stenosis was noted in coronary arteries on catheterization after screening with positive stress tests. Cerebral atherothrombotic infarction was diagnosed on clinical symptoms, CT, MRI and angiography. Diagnosis of ASO was based on claudication, ankle pressure index below 0.9, and angiography. These atherosclerotic subjects were admitted to the Atherosclerosis, Metabolism and Clinical Nutrition wards between 1992 and 1995. Subjects for the population-based Suita study were recruited from people who had been receiving regular health check-ups between April 1995 and February 1996, in The Department of Preventive Cardiology, NCVC. Non-hospitalized subjects participated on a voluntary basis. Their ages ranged between 30 and 79 years old, and they were postulated to be adequate as control subjects.
2.2. Blood samples and ultracentrifugal procedures Blood samples were obtained in EDTA-containing tubes from the subjects following an overnight fast. Plasma was immediately separated and kept at 4°C. VLDL, IDL, LDL and HDL subfractions were separated by ultracentrifugation using a Beckman TL-100 essentially according to the method described by Hatch Table 1 Patient profiles (mean 9S.E.)
Age (years) Gender (M/F) Chol (mg/dl) TG (mg/dl) HDLC (mg/dl) Apo B (mg/dl)
Healthy controls (n =18)
Patients (n = 2161)
49.2 95.7 7/11 186.2 97.2 77.0 95.9 61.3 93.3 86.0 94.0
57.9 9 0.27 1151/1010 207.5 90.93 130.0 91.4 44.4 90.32 121.3 90.73
Y. Hattori et al. / Atherosclerosis 138 (1998) 289–299
293
Fig. 4. (a) Correlation of LDL-chol values between measured and estimated method for healthy control subjects; (b) Correlation of LDL-apo B values between measured and estimated method for healthy control subjects; (c) Correlation of LDL-chol/LDL-apo B values between measured and estimated for healthy control subjects.
294
Y. Hattori et al. / Atherosclerosis 138 (1998) 289–299
Fig. 4. (Continued)
and Lee [10]. Those with obvious chylomicronemia (floating cake layer) were excluded. Chol and TG were determined by enzymatic method (Kyowa Medix, Tokyo) [11] as described before.
2.3. Apo B determination by the latex method A 10% aqueous suspension of a polystyrene latex (particle size 0.2 mm) was diluted 10-fold with phosphate buffer (pH 7.2). To 10 ml of the aqueous suspension of polystyrene latex, 2 ml of 1 mg/ml solution of the anti-apolipoprotein B specific purified antibody and 0.5 ml of 1 mg/ml solution of 1-ethyl-3 (3-dimethylaminopropyl)-carbodiimide-hydrochloride were added and the mixture reacted for 4 h at 5°C. The mixture was centrifuged at 10000 rev./min for 1 h to remove the excess anti-apolipoprotein B specific purified antibody. The antibody-coated polystyrene latex thus obtained was suspended in 50 ml of a glycine buffer (pH 8.2) containing 0.3% bovine serum albumin. To this aliquot (250 ml), 220 ml of 0.1 M glycine buffer (pH 8.6) was added, then 30 ml specimen and the mixture stirred. The turbidity and the transmission at 3.5 min was determined with an automated immunoassay system (EL-1000, Kyowa Medix, Japan). The method is partic-
ularly sensitive and the plasma and LDL were diluted 100-fold and the other subfractions were diluted 30-fold before the assay. CV was between 1.5–2.5%. Since apo B antibody which was raised against human LDL was polychronal, standard curves for VLDL, IDL, LDL, were identical [12]. Although chylomicron behaved in a way similar to LDL, obvious chylomicronemia was excluded from the study. Therefore, both apo B-100 and apo B-48 were detectable by the present method. The latex method is more accurate compared with immunoturbidity method which tends to produce a higher value for samples with high TG [4,6,11,12]. Apo B values determined by the latex method correlate closely with those produced by enzyme immunoassay [13]. In contrast to the commercially available immuno-turbidity method, which shows erroneously higher VLDL-apo B value in hypertriglyceridemia, the latex method gives accurate values for apo B in both VLDL and LDL.
2.4. Mathematical procedures From the thousands of data in lipoprotein subfractions, we noted that a constant relation exists among TG, Chol and apo B content in VLDL, IDL and LDL.
1514 43 46 20 42
674/840 30/13 32/14 13/7 33/9
Gender
22.790.1 24.4 9 0.5** 24.99 0.5** 22.0 90.7 22.4 9 0.4
BMI
Values are mean 9S.E. (mg/dl). By t-test vs control, *PB0.05; **PB0.01.
Control IHD CAD MI A-Th ASO
No of subjects 21090.8 2299 7.0** 2079 6.5 21998.9 21896.1
Chol
HDLC 59.1 9 0.4 47.9 92.1** 43.8 91.9** 51.89 3.6** 49.29 2.3**
TG
112 92.5 172 913.2** 146 99.0* 154 9 20.0 141 910.9*
Table 2 Lipoprotein profiles for control and macroangiopathy (MA: IHD, CVD, ASO)
98.0 90.9 134 9 4.7** 126 9 5.0** 117 9 7.6* 124 9 4.4**
Apo B
17.6 9 0.3 25.6 92.1** 21.6 9 1.4* 22.9 9 2.4* 21.8 9 1.7*
VLDLC
0.149 0.0003 0.14 9 0.003 0.15 90.001 0.15 9 0.009 0.15 90.001
VLDLC/TG
119.4 9 0.8 136.2 9 6.2** 124.0 95.6 126.5 9 7.5* 127.3 95.1**
LDLC
91.89 0.7 111.49 4.1* 105.594.4 96.89 6.4 104.093.7*
LDLB
1.39 90.01 1.26 90.06* 1.219 0.04** 1.36 90.07 1.25 90.05*
LDLC/LDLB
Y. Hattori et al. / Atherosclerosis 138 (1998) 289–299 295
Y. Hattori et al. / Atherosclerosis 138 (1998) 289–299
296
Fig. 5. New classification of hyperlipoproteinemia including apo B. Table 3 Association between lipoprotein phenotype and macroangiopathy Lipoprotein phenotypes
Hyperapo B (all) Hypertriglyceridemic Normotriglyceridemic Type IIa Type IIb Type IV Normal Hypoalphaproteinemia Total Combined low-HDLC
Control
275 89 186 566 187 101 371 14
(18.2) (5.9) (12.3) (37.4) (12.3) (6.7) (24.5) (0.9)
1514 107 (7.1)
IHD CAD
MI
10 7 8 10 19 1 3 0
16 9 7 10 8 4 7 1
(23.3) (16.3)* (18.6) (23.3) (44.2)** (2.3)** (7.0)** (0)**
43 11 (25.6)**
A-Th (34.8)** (19.6)* (15.2) (21.7)* (17.4) (8.7)** (15.2)** (2.2)**
46 21 (45.7)**
3 1 2 3 7 2 4 1
(15.0) (5.0) (10.0) (15.0) (35.0)** (10.0)** (20.0)** (5.0)**
20 5 (25.0)**
ASO 9 6 3 12 12 1 8 0
(21.4) (14.3)* (7.1)** (28.6) (28.6)** (2.4)** (19.0)** (0)**
42 11 (26.2)**
Values are number of subjects with percentage in parentheses. By Chi square analysis, *PB0.05, **PB0.01.
HDL-apo B includes LP(a) and cannot be ignored. Theoretically the following equations are given for total plasma cholesterol and apo B concentration: Chol = VLDLC + IDLC +LDLC +HDLC; apo B= VLDLB + IDLB +LDLB +HDLB; where subscript C is Chol, and subscript B is apo B. The following formulae are assumed with constant a and b for Chol and TG: VLDLC +IDLC = a(Chol −HDLC) +bTG; VLDLB +IDLB + HDLB =a(Chol −HDLC) +bTG. Regression analyses using the actual data have been carried out using SAS [14,15] (software package for statistical analyses). LDLC and LDLB were obtained by
subtracting the above values from plasma Chol other than HDLC and apo B.
3. Results
3.1. Estimation of LDL-chol Regression analyses using the equation shown in Section 2.4 have been carried out with SAS and have produced the following formula: VLDLC + IDLC = 0.06(Chol− HDLC)+ 0.19TG
(1)
Y. Hattori et al. / Atherosclerosis 138 (1998) 289–299
The correlation between estimated (VLDLC +IDLC) and measured values is 0.71. The following formula was then obtained after subtracting the above value and HDLC from total chol:
297
0.94Chol− 0.94HDLC − 0.19TG LDLC = LDLB apo B− 0.09Chol+ 0.09HDLC − 0.08TG A significant correlation was noted between measured and estimated values (r= 0.67, PB 0.001) (Fig. 3).
LDLC = Chol− HDLC −VLDLC −IDLC = 0.94Chol−0.94HDLC −0.19TG
(2)
The value LDLC was calculated for each subject according to Eq. (2). Plots of each LDLC calculated by this method versus that obtained with actual ultracentrifugation are shown (Fig. 1A). A good correlation (r= 0.86) between estimated and measured LDLC was observed. A somewhat better, but similar correlation was noted by the present equation compared with Friedewald’s [16], although the present value for LDLC does not theoretically include IDLC (Fig. 1B).
3.2. Estimation of VLDL-chol The following formula is assumed for VLDLC, with constant a, b for Chol and TG. VLDLC =a(Chol− HDLC) +bTG Regression analyses on this model have been carried out using the SAS program, which derive the following formula: VLDLC = − 0.01(Chol−HDLC) +0.16TG
(3)
The value VLDLC was calculated for each subject according to Eq. (3). A significant correlation between estimated VLDLC and the measured value was noted with r = 0.81 (PB0.0001, data not shown). VLDL-chol is valuable for the diagnosis of type III hyperlipidemia (VLDL-chol/TG \ 0.3).
3.3. Estimation of LDL-apo B Using individual data, regression analyses by SAS gave the following equation: VLDLB +IDLB + HDLB = 0.09(Chol−HDLC) +0.08TG A significant correlation between estimated (VLDLB +IDLB + HDLB) and measured value was noted with r =0.61 (P B0.001, Fig. not shown). Finally, LDLB was obtained as follows:
3.4. Calculated lipoprotein analyses in control healthy subjects LDL-chol in control healthy subjects falls between 50 and 130 excluding two cases. A high correlation (r= 0.95) was observed between measured and estimated values (Fig. 4A). Calculated LDL-apo B ranges from 40 to 90 mg/dl (90% percentile) with good correlation with the actually measured values (Fig. 4B). The ratio of the above estimated LDL-chol/LDL-apo B in the control subjects is shown in Fig. 4C. Normal values range from 1.2 to 2.3 (90% percentile). This ratio reflects the size of LDL and the preponderance of small dense particles is postulated when the ratio falls below 1.2.
3.5. Calculated lipoprotein analyses in subjects with atherosclerotic cardio6ascular diseases and population-based Suita study In CAD, in addition to the elevation of serum Chol, TG and apo B with low HDL-chol, increased LDL-apo B and LDL-chol and lower ratio of LDL-chol/LDL-apo B were noted, indicating the increased number of smaller size of LDL (Table 2). In MI the same abnormality was noted except that the elevation of LDL-apo B was not statistically significant when compared with normal LDL. In ASO hyperapobetalipoproteinemia was observed, since normal LDL-chol and increased LDL-apo B and lowered LDL-chol/LDL-apo B ratio were noted. In atherothrombotic cerebral infarction, elevation of serum apo B with low HDL-chol and increased VLDLchol and LDL-chol were noted. Using the above lipoprotein profiles and new classification including apo B (Fig. 5), the above subjects with atherosclerosis were classified into the following phenotypes (Table 3). Hyperapobetalipoproteinemia with high TG was three times more frequent in CAD, MI and ASO over Suita population. Type IIb was 2.5–3.5 times more frequent in CAD, A-Th and ASO while the other phenotypes were less remarkable and normolipidemia was significantly less in CAD, A-Th and ASO. Low HDL-chol with combined other phenotypes was 3- to 7-fold more prevalent in cardiovascular diseases.
LDLB = apo B− VLDLB −IDLB −HDLB = apo B− 0.09Chol +0.09HDLC −0.08TG The close correlation (r = 0.87) between estimated and actually measured LDLB values is shown in Fig. 2. LDL-chol/LDL-apo B can be calculated from the above formulae as follows:
4. Discussion We developed estimated formulae for LDL-chol, LDL-apo B, VLDL-chol and LDL-chol/LDL-apo B using plasma Chol, TG, HDL-chol and apo B. Ultracen-
298
Y. Hattori et al. / Atherosclerosis 138 (1998) 289–299
trifugal data from 2179 subjects were used for obtaining the above formula. The equation has been developed under the assumption that VLDLC +IDLC is the function of (Chol−HDLC) and TG. Then LDL was calculated by subtracting VLDLC +IDLC and HDLC from total chol. The value thus obtained theoretically does not contain IDL while Friedewald’s formula for LDLchol (Chol−HDL-chol− 0.2TG) contains IDL-chol. Actually, Friedewald’s formula gives somewhat higher results than the present values (Fig. 1A,B). LDL-apo B has been calculated based on the assumption that VLDLB +IDLB + HDLB is the function of (Chol − HDL-chol) and TG and the correlation with the actual value was highly significant (r =0.87, P B 0.001). LDLapo B value is necessary for the diagnosis of hyperapobetalipoproteinemia (HBL), which is the most common lipoprotein disorder in cardiovascular diseases [17]. A small dense LDL, which shows a low ratio of LDLC/ LDLB is also a disorder characteristic of HBL. This disorder is often observed in diabetes [4], obesity [5], HT [18], and hyperlipidemia as well as in cardiovascular diseases [8,19]. The difference between the estimated and measured values divided by the actual mean has been plotted against measured values. The percent samples which fall between −0.3 0.3 is 66.7% for LDLchol, 84.3% for LDL-apo B and 86.0% for LDL-chol/LDL-apo B, respectively. The above values for LDL-chol (66.7%) and the others were higher than the Friedewald’s formula (51.0%). These data suggest that although the estimated formula has some limitations, it can be used for the analysis of lipoprotein disorder. Tests of the significance of b for the regression equations were all at the 1% level. A new classification of hyperlipidemia using apo B has been advocated by Kwiterovich et al. [9]. Using the present formula for LDLC and LDLB, this new classification can be applied. Without performing ultracentrifugal separation, additional information is obtained for the quantitative and qualitative alteration of LDL, which is considered to be a risk for cardiovascular diseases. This classification is more suitable than the classification of WHO since HBL was noted in 20 – 40% of patients with cardiovascular disease. The strict definition of HBL is normal LDL-chol with elevated LDL-apo B, which is often observed in myocardial infarction [8] as well as in acute coronary syndrome and characteristic disorders shown in obesity, diabetes and hypertension [18]. In fact, HBD with high TG was 2.5 – 3 times more frequent in CAD, MI and ASO. The characteristic feature of small dense LDL, lower ratios of LDL-chol/LDL-apo B were observed in these three groups. The present formulae are valuable for the detailed analysis of lipoprotein disorder since additional information regarding qualitative as well as quantitative properties of lipoprotein can be obtained, although there is the limitation that this is an estimation, not the actual value.
Acknowledgements This study was supported in part by Special Coordination Funds for promoting Science and Technology (Encouragement System of COE) and from a grant for Scientific Research Expenses for Health and Welfare Programs (Clinical Treatment of Diabetes Mellitus, Akanuma).
References [1] Castelli WP, Garrison RJ, Wilson PWP, Abbott RD, Kalousdian S, Kannel WB. Incidence of coronary heart disease and lipoprotein cholesterol levels. The Framingham Heart Study 1986. J Am Med Assoc 1986;256:2835. [2] Zhao X-Q, Brown BG, Hillger L, Sacco D, Bisson B, Fisher L, Albers JJ. Effects of intensive lipid-lowering therapy on the coronary arteries of asymptomatic subjects with elevated apolipoprotein B. Circulation 1993;88:2744. [3] Sniderman AD, Wolfson C, Teng B, Franklin FA, Bachorik PS, Kwiterovich PO Jr. Association of hyperapobetalipoproteinemia with endogenous hypertriglyceridemia and atherosclerosis. Ann Intern Med 1982;97:833. [4] Harano Y, Kageyama A, Nakao Y, Hara Y, Suzuki M, Sato A, Ikebuchi M, Shinozaki K, Tsushima M. Quantitative and qualitative derangement of apolipoprotein B-containing lipoproteins as a risk factor for diabetic macroangiopathy in nonobese NIDDM subjects. Diabetes 1996;45(Suppl):S31. [5] Lanhdenpera¨ S, Syva¨nne M, Kahri J, Taskinen M-R. Regulation of low-density lipoprotein particle size distribution in NIDDM and coronary disease — importance of serum triglycerides. Diabetologia 1996;39:453. [6] Harano Y, Nakao Y, Kageyama A, Suzuki M, Hirose J, Asakura Y, Sato A, Komatsu R, Tsusima M, Yamamoto A. A contribution of hyperinsulinemia and hyperglycemia to lipoprotein disorder in obesity. In: Oomura Y, Tarui S, Inoue S, Shimazu T, editors. Progress in Obesity Research. London: Libby, 1990:299 – 302. [7] Schwandt P. The triglyceride controversy — a review of the data. Eur Heart J 1990;11(Suppl H):43. [8] Yokota C, Nonogi H, Miyazaki S, Goto Y, Haze Y, Hara Y, Shinozaki K, Ikebuchi M, Suzuki M, Harano Y. Lipoprotein analyses in patients with stable angina and acute coronary syndrome. Int J Cardiol 1996;57:161. [9] Kwiterovich PO Jr, Coresh J, Bachorik PS. Prevalence of hyperapobetalipoproteinemia and other lipoprotein phenotypes in men (aged 5 50 years) and women (aged 560 years) with coronary artery disease. Am J Cardiol 1993;71:631. [10] Hatch FT, Lee RS. Practical method for plasma lipoprotein analysis. Adv Lipid Res 1968;6:1. [11] Ryomoto KI, Suzuki M, Tsushima M, Yamamoto A, Harano Y. Effects of simvastatin on the number and composition of apoprotein B-containing lipoproteins in hypercholesterolemia: analysis of apoprotein B in each lipoprotein fraction by highly sensitive latex method. Endocr J 1996;43:469. [12] Harano Y. Highly sensitive apo B assay and its clinical significance. Jpn J Clin Pathol 1990;38:861. [13] Harano Y, Kojima H, Nakano T, Harada M, Kashiwagi A, Nakajima T, Hidaka TH, Ohtsuki T, Suzuki T, Tamura A, Fujii T, Nishimura T, Ohtaka T, Shigeta Y. Homozygous hypobetalipoproteinemia with spared chylomicron formation. Metabolism 1989;38:1.
Y. Hattori et al. / Atherosclerosis 138 (1998) 289–299 [14] Freund J, Little RC. SAS System for Regression. 2nd ed. Cary, NC: SAS Institute, 1992. [15] Schlotzhauer SD, Little RC. SAS System for Elementary Statistical Analysis. Cary, NC: SAS Institute, 1991. [16] Friedewald T, Levy RI, Fredrickson DS. Estimation of the concentration of low density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem 1972;18:499. [17] Kwiterovich PO Jr, Motevalli M, Woodford FP, Davignon J, Sniderman A, editors. Hyperapo B—An Update. Atherosclero-
.
299
sis X. Amsterdam: Elsevier, 1995:966 – 971. [18] Ikeda K, Suzuki M, Ikebuchi M, Hara Y, Tsushima M, Yamamoto A, Harano Y. Hyperbetalipoproteinemia with small low-density lipoprotein, a characteristic disorder of lipoprotein in essential hypertension. J Diabetes Complications 1995;9:227. [19] Sato A, Kanazawa A, Hasegawa M, Asakura Y, Yokota C, Suzuki M, Tusima M, Kuriyama Y, Sawada T, Yamaguchi T, Kobayashi M, Harano Y. Lipoprotein disorder in brain infarction and hemorrhage. Int J Clin Lab Res 1998 (in press).
.