without diabetes mellitus

Kidney International, Vol. 44 (1993), PP. 1062—1070

Apolipoprotein(a) phenotypes and serum lipoprotein(a) levels in maintenance hemodialysis patients with/without diabetes mellitu s KYOKO H!pATA, SHuIcH! KIKucHI, KEIJIR0 SAKU, SHIR0 JIM!, Bo ZHANG, SETSUYA NAIT0, HIDE0 HAMAGUCHI, and KIKuo ARAKAWA Departments of Internal Medicine and Pathology, Fukuoka University School of Medicine, Fukuoka, Japan, and Department of Medical Genetics, Institute of Basic Medical Sciences, University of Tsukuba, Tsukuba, Japan

Apolipoprotein(a) phenotypes and serum lipoprotein(a) levels in maintenance hemodialysis patients with/without diabetes mellitus. We studied the quantitative and qualitative characteristics of lipoprotein(a) [Lp(a)]

as a function of apolipoprotein(a) [apo(a)1 phenotypes in 152 patients (123 males, 29 females) undergoing maintenance hemodialysis (HD) with or without diabetes mellitus (DM), in 101 patients with diabetes mellitus without hemodialysis (58 males, 43 females), and in 421 normal controls (333 males, 88 females). Serum Lp(a) levels were significantly (P < 0.01) higher in patients than in controls (26.2 18.3 mg/dl in HD with DM, 26.4 22.0 mg/dl in HD without DM, 27.1 27.3 mg/dl in DM without HD, and 14.9 13.7 mg/dl in controls, respectively).

levels of Lp(a) are associated with an increased risk for coronary ieart disease (CHD) [6—9], and Lp(a) is assumed to be an indepcndent risk factor for CHD. Recent studies have shown that higher concentrations of plasma Lp(a) can also be seen in

patients with diabetes mellitus (DM) [10, 11], or who are undergoing hemodialysis (HD) or continuous pentoneal dialysis

[12—15]. We previously reported that serum Lp(a) levels in hemodialysis patients were significantly higher, and that the frequency distributions of serum Lp(a) levels in these patients Apo(a) phenotyping was performed by a sensitive, high resolution differed from those in a normal control group [16]. technique using SDS-agarose/gradient (3 to 6%) PAGE. In normal Plasma Lp(a) levels in humans vary from less than 1 mgldl to controls, the molecular weights of apo(a) isoforms were inversely correlated with plasma Lp(a) levels, and the same tendency was found in patients who were undergoing hemodialysis and/or who had diabetes mellitus. We assumed the differences in apo(a) phenotypes detectable with our method reflected consecutive differences in molecular weights of apo(a). The results of an analysis of covariance and a least square means comparison indicated that the regression lines between serum Lp(a) levels [log Lp(a)] and apo(a) phenotypes in patient groups were significantly (P < 0.01) elevated for every apo(a) phenotype, as compared to the regression line of the control group. Even after the low molecular weight apo(a) phenotypes (Al-A8) were omitted, the same tendency was observed. However, no differences were observed between the patient groups. The frequency distributions of serum Lp(a) levels in the patient groups were significantly different at every cut off point [10, 20, 30, 40, and 50 mg/dl of Lp(a)1 from those in the control

group, as assessed by chi square statistics. These data indicate that serum Lp(a) levels are controlled by alleles at the apo(a) locus in patients undergoing hemodialysis, as well as in patients with diabetes mellitus, but that increased Lp(a) levels in hemodialysis and diabetic patients are not regulated by genes encoding low molecular weight of apo(a) isoforms.

more than 100 mg/dl, and these levels are genetically controlled

by apo(a) polymorphism, that is, individuals who carry isoforms with lower molecular weights are more likely to have higher levels of plasma Lp(a) than individuals who carry isoforms with larger molecular weights [17—19]. Since Utermann et al first reported a genetic size polymorphism of six apo(a) isoforms with different molecular weights [17], several other researchers have detected additional apo(a) isoforms [20,

21]. We have recently detected 26 different apo(a) alleles, including a null allele, at the apo(a) locus by a sensitive, high resolution technique using SDS-agarose/gradient PAGE, and have observed a highly significant inverse correlation between serum Lp(a) levels and the size of apo(a) isoforms [22]. Lipid abnormalities have also been observed in patients with renal disease, such as nephrotic syndrome and end-stage renal disease, and in patients with diabetes mellitus [23—27]. Therefore, we determined apo(a) phenotypes, serum Lp(a) levels,

and other lipid and lipoprotein levels, in patients who were undergoing maintenance hemodialysis with/without diabetes Lipoprotein(a) [Lp(a)] is a cholesterol-rich lipoprotein which mellitus, as well as in patients with diabetes mellitus without has a composition similar to that of low density lipoprotein hemodialysis. This paper reports the results of our investigation (LDL). Apolipoprotein(a) [apo(a)], an important constituent of of the quantitative and qualitative characteristics of Lp(a) and Lp(a), is a glycoprotein which shows a high homology to its relationship to the diseased state as a function of apo(a) plasminogen, and is linked to apo B-l00 by disulfide bridges in phenotypes. Lp(a) [1—5]. Many clinical reports have shown that elevated Received for publication March 29, 1993 and in revised form June 22, 1993 Accepted for publication June 24, 1993

© 1993 by the International Society of Nephrology

Methods

Patients One hundred and fifty-two patients (123 males, 29 females, from 23 to 77 years old, 56.1 11.6 years) undergoing maintenance hemodialysis (17 of whom had coronary heart disease), 1062

1063

Hirata et a!: Apolipoprotein(a) phenotypes in hemodialysis

*

101 patients with non-insulin dependent diabetes mellitus (NIDDM) (58 males, 43 females, from 42 to 89 years old, 63.9

70

10.8 years) and 421 normal controls (333 males, 88 females, from 19 to 69 years old) recruited from eight different hospitals, were studied. Controls were selected from individuals who had received an annual medical check-up at three different hospitals. Among the hemodialysis patients, 30 (21 males, 9 females) were receiving maintenance hemodialysis because of diabetic

1

60 —

50

40

nephropathy, 18 (18 males) other diabetes patients were on maintenance hemodialysis but the etiology of renal failure were unknown, and 122 patients (101 males, 21 females) were receiving hemodialysis for other reasons, such as nephrosclerosis (N

20. 10-

= 11), chronic glomerulonephntis (N = 48), renal cell carcinoma (N = 1), polycystic kidney (N = 2), pyelonephritis (N = 3), IgA nephropathy (N = 4), gout (N = I), systemic lupus erythematosus (N = 2), rheumatoid arthritis (N = 1), rapidly Fig. 1. Serum Lp(a) concentrations in patients with hemodialysis progressive glomerulonephntis (N =

(N =

1),

and unknown etiology

48). Although a renal biopsy was not conducted in all patients, the reasons for hemodialysis depended, in part, on a clinical diagnosis. The duration of hemodialysis ranged from 1 to 252 months (66.9 51.3 months). None of the patients were treated with CAPD. The dialysis schedule was four or five hours three times a week. We used a 0.8 to 2.1 m2 surface area hollow-fiber filter [cuprophan membrane (31.7% of the patients), triacetate membrane (13.1%), modified cellulose membrane (14.5%), saponificated cellulose membrane (1.4%), polymethylmethacrylate membrane (PMMA; 23.5%), ethylenevinyl alcohol copolymer membrane (EVA; 5.5%), polysulfon membrane (10.3%)], and bicarbonate dialysate containing sodium 140 mEq/liter, kalium 2.0 mEq/liter, calcium 3.0 mEq/liter, chlorine 110 mEq/liter, acetate 8.0 mEq/liter, carbonic acid 30 mEq/liter, and glucose 100 mg/dl or 150 mg/dl (150 mg/dl was used for insulin-treated diabetic patients). Among the diabetic patients without hemodialysis, 10 patients were receiving insulin treatment, 55 were receiving oral hypoglycemic agents, and 36 patients were receiving single diet therapy. Diagnosis of diabetes was based on intolerance to 75 grams of oral glucose, the use of oral hypoglyceridemic agents, or insulin therapy. Diagnosis of hypertension was based on blood pressure above 160/95 mm Hg more than two times, or the use of anti-hypertensive drugs. Diagnosis of coronary heart disease was based on coronary angiography, electrocardio-

without diabetes mellitus (N = 104), hemodialysis with diabetes mellitus (N = 48), diabetes mellitus without hemodialysis (N = 101), and controls (N = 421). * P < 0.01.

immunoassay (TIA) method [33]. Serum Lp(a) was determined by ELISA using a Tint Eliza Lp(a) kit (Biopool, Sweden) [34]. Urine proteins (albumin) were measured using spot urine ad-

justed for urinary creatinine as described by Takegoshi et a! [11]. Urinary albumin was assayed by an immunoturbidity method using a Riket Albumin kit (Japan Chemipha, Tokyo) [35]. Interassay and intra-assay coefficient variations of all measurements in our laboratory were less than 5.5%. Plasma lipoprotein isolation Venous blood was collected in EDTA tubes and then centri-

fuged at 4°C. The plasma (approximately 1 ml) fraction at density <1.12 g/ml was separated by ultracentrifugation for four

hours (using a Beckman TL-100 Tabletop Ultracentrifuge, TLA-lOO.2 fixed anglerotor), using polycarbonate centrifuge tubes. Tube slicing was performed using a Centrifuge Slicer (Beckman Instruments, Inc., Palo Alto, California, USA).

Apo(a) phenotyping Apo(a) phenotyping was performed in 101 hemodialysis patients without diabetes mellitus, 44 hemodialysis with diabetes gram, or clinical heart attack episodes. While some of the mellitus, 94 diabetes mellitus, and 215 control subjects. Apo(a) patients were treated with anti-hypertensive drugs, neither phenotyping was performed by the method of Kikuchi et al [22]. diuretics nor beta-blockers were used. Therefore, there was Briefly, 2 to 5 d lipoprotein fraction (d < 1.12 g/ml) was mixed minimal possibility of adverse effects from antihypertensive with an equal volume of a sample buffer which contained 0.25 M drugs on lipid metabolism, including that of Lp(a). Hypolipid- Tris-HCI (pH 6.8), 20% sucrose, 10% SDS, 10% 2-mercaptoemic agents were not used in the patient and control groups. ethanol, and 0.02% BPB. For the stacking gel, 1.2% agarose gel containing 0.125 M Tris-HC1 (pH 6.8), 0.2% SDS, and 0.5% Chemical determinations 2-mercaptoethanol was used instead of polyacrylamide gel. For Blood sampl.s were taken before hemodialysis and after the running gel, we used 3 to 6% gradient polyacrylamide slab overnight fasting. Serum total cholesterol (TC) and triglycerides gels containing 0.375 M Tris-HC1 (pH 8.8) and 0.2% SDS. The (TG) were determined by enzymatic methods [28, 29]. The running buffer contained 25 msi Tris, 0.19 M glycine, and 0.2% heparin CaCI2 precipitation method was used to assay serum SDS. A control sample contained the lipoprotein fraction from high-density lipoprotein cholesterol (HDL-C) [30]. Low-density four individuals. Electrophoresis was performed at a constant lipoprotein cholesterol (LDL-C) was calculated using the for- current of 5 mA for three hours and then at 3 mA for an mula of Friedewald, Levy and Fredrickson [31]. HDL2 and additional 18 hours at room temperature. Electrophoretic transHDL3 subfractions were separated by an ultracentrifugation fer of protein from gel to nitrocellulose membrane filters (BA83, method [30, 32]. Serum apo Al, apo All, apo B, apo CII, apo Schleicher & Schull) was performed for two hours at a constant CIII, and apo E levels were determined by the turbidity current of 180 mA with a blotting buffer (7 x Running buffer

1064

0 9

9

CD

9

9

-.4

9

01

39

r9

0)

9 a

0)

9

N)

9

9

a 0

0,

0

Hirata et a!: Apolipoprotein(a) phenotypes in heinodialysis

Fig. 2. Frequency distribution of plasma Lp(a) concentrations in patients with hemodialysis without diabetes mellitus (fl, N = 104), hemodialysis with diabetes mellitus N = 48), diabetes mellitus without

(, hemodialysis (, N = N = 421).

101) and

controls (•,

Table 1. Serum lipids, lipoproteins, and apolipoproteins in patients and controls HD without DM

N

HD with 48

104

TC TG HDL-C LDL-C HDL2-C HDL3-C

apoAl

apo All apo B apo CII apo CIII apo E

39.3

357c 72.lc 13.lc

102.7

31.Oa

168.1

130.6

DM without HD

DM 154.8 125.8

33.1 96.4 20.7

24.0

lO.3c

16.1

3.7k'

107.4

20.6a

25.5 85.0 3.6

59C 24.3c

l.7c

21.6 81.8 3.2

15.7

59C

5.0

1.8c

13.8 95.2

101

41.6c

78,lc

i1.7

29.8c

9,lc 3,3C

21.3c 6,3c

28.3"

215.8 192.0

554C

272.2

Controls 421 178.5 24.3 84.1 30.2

47.2

l4.2c

52.7

130.1

54.3

108.9

13.3 24.1

1l.6c

10.4 3.2

28.8 19.6

44b

35,2 20.6

128.7

30.5"

135.2

32.5 105.6

10.0

29.0 72.4 2.8 8.5

4.3

4.1

1.0

1.8c

4.8

12.7

1,8c

1.8

13.8

4,7

6.8

34.1c

3.lc

l0.8

5,9C

21.3 14.7 1.1 2.3

ap< 0.1 b P < 0.05

P < 0.01

containing 10% methanol). The filter was blocked with 0.2% gelatin in PBS and incubated in 1% BSA in PBS containing 0.1% anti-apo(a) monoclonal antibody. The anti-apo(a) monoclonal antibody was prepared using purified human apo(a) and did not cross react with human plasminogen or other plasma proteins (Nakagawa et al, unpublished observations). Peroxidase-conjugated anti-mouse IgG rabbit serum (0.1% in PBS)

was used as the second antibody, and the patterns were

covariance was used to compare groups after adjusting for confounding factors (age, sex, diseases) and apo(a) phenotypes,

and least square (LS) means were also compared alter log transformation of Lp(a) and after adjusting for apo(a) phenotypes to gain a better understanding of the relationship between the frequency distribution of apo(a) phenotypes and serum Lp(a) levels in diseased and control groups.

Results Serum Lp(a) concentrations in patients with hemodialysis but Statistical procedures without diabetes mellitus (N 104), with hemodialysis and All calculations were performed using the Statistical Analysis diabetes mellitus (N = 48), with diabetes mellitus but without System (SAS). The data are all presented as the mean SD. hemodialysis (N = 101) and controls (N = 421) were 26.4 22.0 Because a Kolmogorow-Smirnov test showed significant devi- mgldl, 26.2 18.3 mgldl, 27.1 27.3 mg/dl, and 14.9 13.7 ation from the normal distribution for some biochemical vari- mgldl, respectively (Fig. 1). The Mann-Whitney U test showed ables, the Mann-Whitney U test was used to compare values in that Lp(a) and log Lp(a) levels were significantly (P < 0.01) the patient groups with those of the controls. P values of less higher in patient groups than in controls, however, no signifithan 0.05 were considered to indicate statistical significance. cant differences were observed between the three different Chi square analysis was used to examine the frequency distri- patient groups. The distributions of serum Lp(a) concentrations bution of Lp(a) levels in patients and controls. Analysis of in the above groups are shown in Figure 2. The median values

visualized using 0.01% DAB and 0.1% H202 in PBS.

1065

Hirata et a!: Apo!ipoprotein(a) phenotypes in hemodialysis M

1

2

M

M

3

4

5

6

7

8

M

9

10

11

M

12

13

14

M

15

16

17

18

19

20

21

22

23

24

Fig. 3. Apo(a) phenolyping using 3 to 6% SDS-a garosel gradient PA GE followed by immunoblot analysis. Lane 1: A12A14, Lane 3: A15A19, Lane 4: A3A15, Lane 5: A14A19, Lane 7: A17A20, Lane 9: A18A25, Lane 10: A18A19, Lane 12: A6A21, Lane 13: A3A19, Lane 15: A12, Lane 16: A15A18, Lane 18: A16A18, Lane 19: A9, Lane 21: A18, Lane 22: A8A16, Lane 24: A16A18, Lanes 2,6,8, 11, 14, 17, 20, and 23: a marker mixture (M) containing apo(a) isoforms AlO, A12, A14, A16, A18 and A21.

and ranges of Lp(a) concentrations in hemodialysis without apo(a) isoforms, because the smaller band was well correlated DM, hemodialysis with DM, DM without hemodialysis, and with Lp(a) levels in double-banded phenotypes [22]. In normal controls were 21.5 (0.5 to 113) mg/dl, 22.5 (0.5 to 98) mg/dl, 18.0 controls, apo(a) isoforms were a strong detennining factor of (ito 132) mgldl, and 10.9 (0.4 to 112) mgldl, respectively. In our serum Lp(a) levels (Fig. 4D), that is, an increase in the present study, diabetic or hemodialysis patients often had other molecular weight was significantly (P < 0.01) correlated with atherosclerotic factors, that is, hypertension, coronary heart decreased levels of serum Lp(a). In patients undergoing hemo-

disease, hyperlipoproteinemias, etc. Therefore an analysis of dialysis with/without DM, or DM without hemodialysis, a covariance was performed after making adjustments for age, similar tendency (P < 0.01) was observed (Fig. 4A, B, C). sex, presence of hypertension, coronary heart disease, diabetes In order to assess the percentage changes in Lp(a) at any mellitus, hyperlipoproteinemias and hemodialysis. Hemodialy- isoform more accurately, serum Lp(a) levels were transformed sis treatment was found to most significantly (P < 0.01) into logarithms, since log transformation should provide the contribute to the elevated serum Lp(a) levels, although the R2 percent changes of Lp(a) at each isoform. The regression lines value was low (R2 = 0.04). based on log Lp(a) are presented in Figure 5. We used an Serum lipid and lipoprotein levels were also determined in the four different groups (Table 1). Serum TC levels in hemodialysis without DM and with DM were 168.1 35.7 mgldl and 154.8 41.6 mgldl, respectively, which were lower (P < 0.01) than that

analysis of covariance, and least square (LS) means were

in the control group (178.5 24.3 mgldl). Serum TC in DM without hemodialysis was 215.8 55.4 mgldl, which was

we used the data from A9 to A20 since there were fewer patients in other phenotypes. Even when an analysis of co-

significantly higher than that in controls. Serum HDL-C (39.3 14.2 mg/dl, 52.7 13.3 13.1 mgldl, 33.1 11.7 mg/dl, 47.2 mg/dl, respectively), HDL2-C (24.0 10.3 mgldl, 20.7 9.1 mgldl, 28.8 11.6 mgldl, 35.2 10.4 mgldl, respectively), and HDL3-C (16.1 3.7 mg!dI, 13.8 3.3 mg/dl, 19.6 4.4 mg/dl, 20.6 3.2 mgldl, respectively) were significantly lower (P < 0.05) in the patientgroups. Serum TO (130.6 72.1 mgldl, 125.8 272.2 mg/dl, 84.1 78.1 mgldl, 192.0 302 mg/dl, respectively), apo B (85.0 24.3 mgldl, 81.8 28.3 mg/dl, 105.6 34.1 mgldl, 72.4 14.7 mg/dl, respectively), apo CII (3.6 1.7 1.8 mg/dl, 4.8 3.1 mg/dl, 2.8 1.1 mg/dl, mg/dl, 3.2 respectively), apo CIII (15.7 5.9 mgldl, 12.7 1.8 mg/dl, 13.8 10.8 mgldl, 8.5 2.3 mg/dl, respectively) and apo E (5.0 1.8 mg/dl, 4.7 1.8 mg/dl, 6.8 5.9 mg/dl, 4.1 1.0 mg/dl, respectively) levels were significantly (P < 0.05) higher in the patient groups.

compared after adjusting for apo(a) phenotypes. The lines of the diseased groups were significantly (P < 0.01) higher than that of

the control group. In the group of hemodialysis with diabetes, variance was performed between A7 to A25 or A9 to A25, after

omitting the low molecular weight isoproteins (Al-A6, or to A8), the regression lines of the control and diseased groups

were significantly different. Furthermore, we tentatively grouped these 25 different apo(a) phenotypes into three different subgroups (A2-A9, AlO-A16, and A17-A25) and compared their serum Lp(a) levels. Serum Lp(a) levels of these subgroups in the control were 42.0 24.6 mg/dl, 17.7 9.6 mg/dl, 8.6 5.9 mgldl, respectively, which were all significantly (P < 0.05) less than those in the diseased groups (Table 2).

Duration of hemodialysis versus Lp(a) concentration is shown in Figure 6A. We divided the patients into three groups based on the duration of hemodialysis: <12 months, 12 to 60 months and >60 months. However, no significant differences

were observed between the three groups. Whether or not In our investigation, 25 different types of apo(a) isoforms diabetes was controlled (HbAlc > 7.0% or 7.0%; Figure 6B)

were identified by immunoblot analysis following SDS-agarose/ did not affect the correlation between serum Lp(a) levels and gradient (3 to 6%) PAGE [22]. An example of apo(a) phenotyp- apo(a) phenotypes. For the assessment of diabetic nephropaing is shown in Figure 3. A marker mixture (M) containing thy, we divided the patients into three groups with proteinuria, apo(a) polymorphs AlO, A12, A14, Al6, A18, and A21 was <30 mglg creatinine, 30 to 300 mg/g creatinine, and >300 mglg applied to each lane as shown in Figure 3, and the phenotyping creatinine. After adjusting for apo(a) phenotypes, no significant was performed as described above. Approximately 75% of the differences were observed between the three groups (Fig. 6C). subjects had a double band of apo(a), while the remaining Lp(a) frequency distributions with cut off points at 10, 20, 30, subjects had a single band. When the subject had a double band, 40, and 50 mg/dl of Lp(a) in hemodialysis patients with or the smaller band was used for the expression of phenotypes in without DM and in diabetes mellitus were assessed by the analysis of the relationship between serum Lp(a) levels and statistics (Table 3). The frequency distributions of Lp(a) in the

1066

Hirata et a!: Apolipoprotein(a) phenotypes in hemodialysis

150

3

A

100

DM

2

without DM

50

00

0 150

—ii 0)

D

'''I,,,, -VI

-j !!D ,,,,iY-T—ç—-i,p-— 0

0

Controls DM without HO

0

B —1

I

100

0

50

0 0

A

A

A

1

5

10

0

I

I

91

A

A 25

20

Apo(a) isoform phenotype

0

Go a 8°o

A 15

Fig. 5. The regression lines of the correlation of serum Lp(a) concen-

trations and apo(a) polymorphism in patients with (A) hemodialysis

without diabetes mellitus (y = 2.2732 — 6.8018e — 2x, R2 = 0.375), (B) hemodialysis with diabetes mellitus (y = 2.1 747 — 5.3797e — 2x, R2 =

On

0.270), (C) diabetes mellitus without hemodialysis (y = 2.2261 —

150

0.

6.7825e — 2x, R2 = 0.371), and(D)controls(y = 1.9628 — 6.0655e — 2x,

C

R2 = 0.433).

8 0

00

100

0 0 0

50

0

Table 2. Serum Lp(a) levels of three different subgroups in patients and controls

0

0

008

8

8

A1-A9

Controls

lIIIlI 9111

N = 215 HD without DM

150

apo(a) isoform Al0-A16

0

N = 101

0

42.0 24.6 N = 16 (7.4%) 84.5

17.7

.

100

DM without HD

N = 94

8.6

5.9

N 99 (46.0%) N = 100(46.5%) 28.7

189b

143b

13.6

N 8 (7.9%)

N = 52 (51.5%) N = 41(40.6%)

N = 0 (0.0%)

N = 23 (52.3%) N = 21(47.7%)

HD with DM

N = 44

9.6

A17-A25

31.3 76.3

41.Ol

27.5

203b 17.5"

19.1

12.7

124b 11.5k

N = 12 (12.8%) N = 44 (46.8%) N = 38 (40.4%)

ap 0.05 .

50 0

. . S

bP

.

.

•••si . .s •SIIssI . hI

< 0.01, compared with control

S

• •ill liI,i. .

AAAAAAAAAAAAAAAAAAAAAAAAA 1 2 34 5 67 8910111213141516171819202122232425

Apo(a) isoform phenotype Fig. 4. Correlation of serum Lp(a) concentrations and apo(a) polymorphism in patients with (A) hemodialysis without diabetes mellitus (N = 101), (B) hemodialysis with diabetes mellitus (N = 44), (C) diabetes mellitus without hemodialysis (N = 94), and (D) controls (N = 215).

and other atherosclerotic diseases [6—9, 36], as well as with the diseased states mentioned above [10—12, 16]. Atherogenic lipid

abnormalities are also likely to be present under such conditions (Table 1). Thus, the combination of hyperlipoproteinemia and elevated Lp(a) levels may play a role in the progression of

atherosclerosis. Our study demonstrated that serum Lp(a) levels were elevated in patients undergoing hemodialysis with! without diabetes and in simple diabetic patients. There is little debate regarding whether elevated Lp(a) levels are associated with atherosclerotic diseases, but it is unclear

whether serum Lp(a) levels are also increased in diabetic three patient groups were significantly different from that in the control group at every cut off point used.

patients [37—39]. Haffner et al have reported that Lp(a) levels did not increase in NIDDM patients [39], while we observed higher Lp(a) levels in patients with diabetes mellitus (Fig. 1). Discussion This difference may be due to two factors: (1) the patients in the Accelerated atherosclerosis and cardiovascular disease are present study were not selected from a well controlled populafrequently observed in patients with end-stage renal disease or tion, as in the study by Haffner et al, and (2) many of our diabetes mellitus [6, 10, 11, 14—16]. Elevated serum Lp(a) levels diabetic patients had other atherosclerotic conditions, such as have been associated with premature coronary heart disease coronary heart disease (10.7%), hyperlipoproteinemias (52.4%),

Hirata et a!: Apolipoprotein(a) phenotypes in hemodialysis

150

1067

A

. 100

•

a

a

•

0 0

•

00

150

o.

•

. lI1I.

0

-

0

B

•U

100

I

• ••

•

50

•U

• B

50

••

.

.

S

-- ii_j1

•

•

0—

I

150

0 S

100

0

0 Fig. 6. Correlation of serum Lp(a) concentrations and apo(a) polymorphism in patients with: (A) hemodialysis: the patients

o

o

50

•

• o

.

0

AAAAAA

123456

e

were divided into three groups based on the

o• %

g•

c

duration of hemodialysis () <12 months, (I)

o

• I$% AAAAAAAAAAAAAAAA AAA 7 8 9 101112131415161718 19 2021 22232425 Apo(a) isoform phenotype

12 to 60 months, and (0) >60 months; (B) diabetes mellitus: the patients were divided into two groups of controlled DM or not controlled DM as assessed by HbAlc (•) 7.0%,or (0) >7.0%; (C) diabetes mellitus: the patients were divided into three groups with proteinuria by (0) <30 mg/g creatinine, (•) 30 to 300 mglg creatinine, and (0) >300 mg/g creatznlne.

hypertension (53.4%), etc. Thus, elevated LDL-C levels in lipoproteinemias, and hemodialysis, an analysis of covariance diabetic patients (Table 1) may have produced elevated Lp(a) showed that hemodialysis treatment, rather than diabetes, most levels, as suggested by Schernthaner et al [10].

significantly (P < 0.01) contributed to the elevated serum Lp(a)

However, after adjusting for age, sex, and presence of levels, although the R2 value was low (R2 = 0.04). In addition, hypertension, coronary heart disease, diabetes mellitus, hyper- since the mean duration of hemodialysis in our study was

1068

Hirata et al: Apolipoprotein(a) phenotypes in hernodialysis

Table 3. Frequency distributions of serum Lp(a) levels of five different cut off points in patients and control

Total

>10

Lp(a) lO (mg/dl) N (%)

N (%)

24 (22.9) 8 (16.3) 32 (30.8) 235 (56.9)

81 (77.1)a 41(83,7)8 72 (69.2)8 178 (43.1)

Total

N

Lp(a) 20 (mg/dl) N(%)

>20 N(%)

105

51(48.6) 23 (46.9) 55 (52.9) 326 (78.9)

54 (51.4)8 26 (53.1)8 49 (47.1)8 87 (21.1)

Total

Lp(a) 30 (mgldl) N(%)

>30 N(%)

DM

105

HD + DM

49

72 (68.6) 34 (69.4)

104

71(68.3)

33 (31.4)a 15 (30.6)8 33(31.7)8 35 (8.5)

N

HD - DM

105

HD + DM

49

DM Controls

HD — DM

HD + DM DM Controls

104 413

49 104

413

N

LID —

DM Controls

413

378 (91.5)

Total

Lp(a) 40 (mgldl)

N(%)

>40 N(%)

90 (85.7) 39 (80.0) 82 (78.9) 398 (96.4)

15 (14.3)a 10 (20.0)a 22 (21,1)8 15 (3.6)

Lp(a) 50 (mgldl) N(%)

>50 N(%)

N HD — DM

105

HD + DM

49

DM Controls

104

413

Total

N HD — DM

105

LID + DM DM Controls

94 (89.5) 44 (89.8)

11 (10.5)a

49 104 413

91(87.5)

13 (12,5)8

5 (10.2)a

404 (97.8) 9 (2.2) a p < 0.05 (vs. controls), (HD-DM:HD without DM, HD + DM:HD with DM, DM:DM without HD

approximately five years the patient population may have been skewed; the presence of diabetes mellitus and coronary heart disease at the initiation of hemodialysis may affect the mortality of such patients. However, Takegoshi et al [11] showed that Lp(a) levels are significantly higher in patients with diabetes

mellitus and chronic renal failure than in patients without chronic renal failure. A recent study by Kandoussi et al [15]

Low albuminemia in nephrotic syndrome triggers increased hepatic production of proteins, including lipoproteins [16, 23— 26]. However, neither the hemodialysis patients nor the patients with diabetes mellitus in the present study showed low serum albuminemia. Serum Lp(a) levels in patients with diabe-

tes with or without nephropathy (Fig. 6C) also showed no significant differences. Apparently, renal disorder is not involved in increasing Lp(a) levels in diabetic patients, and further, the presence of diabetes does not affect serum Lp(a) levels in patients undergoing hemodialysis (Fig. 5).

It has been shown that the apo(a) gene could account for

more than 90% of the differences between serum Lp(a) concen-

trations in different individuals within a pedigree [19]. An inverse correlation between the size of apo(a) isoforms and the concentrations of Lp(a) has been reported in population and family studies [17—19, 22]. We used an analysis of covariance, assuming that differences in the apo(a) phenotypes (Al to A25) reflect consecutive differences in molecular weights of apo(a), and found that (Fig. 5) the regression lines of patient and control

groups were almost parallel, that is, the regression line in diabetic patients and hemodialysis patients with/without diabetes mellitus presented elevated Lp(a) [log Lp(a)] levels, and serum Lp(a) [log Lp(a)] levels in such patients were significantly (P < 0.01) higher than those in controls for every apo(a) phenotype, after adjusting for confounding apo(a) isoforms. A similar analysis was performed in the three different patient groups, but no statistically significant difference was observed. The distribution of apo(a) isoforms and serum Lp(a) levels in hemodialysis patients with diabetes seemed to differ from those in other diseased groups, perhaps because of the rather small number of patients studied, although tendencies similar to those in the control group were also observed, Even after omitting the low molecular weight apo(a) phenotypes (Al to A8), the same tendency was observed in other two diseased groups. Since we previously reported that the average molecular difference between adjacent apo(a) isoforms was estimated to be 18.4 kDa from A5 to A24 [22], the molecular weight differences from Al to A4 were rather larger, which may have slightly biased the regression lines. However, this should not have affected the

percent changes of Lp(a) between the control and diseased groups, because Lp(a) was transformed into a logarithm. The apo(a) isoforms were again arbitrarily divided into three different groups (Al-A9, Al0-A16 and A17-A25), and the Lp(a) levels

in these three groups were significantly (P < 0.01) higher in patient groups than in controls (Table 2), The differences

observed between the regression lines of the control and

measured plasma Lp(a) levels in predialysis and hemodialysis diseased groups in Figure 5 essentially matched the Lp(a) patients, and in patients on CAPD. They found that the Lp(a) frequency distribution shown in Figure 4 and Tables 2 and 3. All levels in the former two groups were similar and greater, of these results indicate that the increases in serum Lp(a) levels respectively, than those in controls and CAPD-treated patients. in patients undergoing hemodialysis and in diabetic patients In addition, no difference was observed between controls and were not linked to any particular low molecular weight apo(a) CAPD patients. These findings suggest that CAPD treatment isoform, and it may support the idea that there is no difference rather than hemodialysis may affect apo(a) metabolism. On the in the genetic expression of apo(a) in hemodialysis and diabetic other hand, Haffner et al [14] reported that the type of therapy patients as compared to normal populations. These findings used to treat renal failure does not influence Lp(a) levels. differ from those observed in patients with intermittent claudiNeither study adjusted for apo(a) phenotyping. In our protocol cation, where the elevated Lp(a) levels were due to a preponit was difficult to determine the predialysis serum Lp(a) levels of derance of low-molecular weight apo(a) isoforms [36]. Lp(a) in vivo kinetic studies have led to a better understandso many patients since they were already undergoing hemodialysis. Therefore, chronic renal failure during its early phase ing of the mechanism of low and high serum Lp(a) levels, although to date such studies have only been performed in may hold the key to increase serum Lp(a) levels.

Hirata et al: Apolipoprotein(a) phenotypes in hemodialysis

familial hypercholesterolemia, sporadic hyperlipoproteinemia and normolipidemic subjects [40, 41]. Each condition studied showed similar fractional catabolic rates for apo(a). This suggests that synthesis, rather than the catabolic pathway, may play a key role in setting serum Lp(a) levels. On the other hand, a recent study by Portman et al [42] demonstrated an acquired defect in LDL receptor function in lymphocytes from uremic patients. Therefore, if Lp(a) can be specifically cleared by LDL receptors [40, 43], then increased Lp(a) levels in hemodialysis patients may result from the diminished Lp(a) catabolic pathway. However, this would only explain the increased Lp(a)

levels in hemodialysis patients, and not those in diabetic

patients. In the present study we demonstrated that there was no difference between the serum Lp(a) levels of diabetic patients with or without nephropathy (proteinuria), although microalbuminuria itself could be an independent risk factor for cardiovascular disease [44]. Although present, urinary loss of

Lp(a) did not affect catabolic rates of Lp(a). Furthermore, neither the control of diabetes as assessed by HbAlc [37] nor proteinuria affected serum Lp(a) levels (Fig. 6, B and C). These results strongly suggest that Lp(a) synthesis may be increased

with hemodialysis and diabetes mellitus, regardless of the apo(a) phenotypes. It would be interesting to determine the kinetic parameters of Lp(a) from various apo(a) alleles in

1069

3. MBwu AD, DURRINGTON PN: Lipoprotein(a): Structure, properties and possible involvement in thrombogenesis and atherogenesis. Atherosclerosis 85:1—14, 1990 4. MCLEAN JW, TOMLINSON JE, KUANG WJ, EATON DL, CHEN EY,

FLESS GM, SCANU AM, LAWN RM: cDNA sequence of human apolipoprotein(a) is homologue to plasminogen. Nature 330:132— 137, 1987

5. UTERMANN 0: The mysteries of lipoprotein(a). Science 246:904910, 1989 6. DAHLEN OH, GUYTON JR, ATTAR M, FARMER JA, KAUTZ JA, GOTTO AM JR: Association of levels of lipoprotein Lp(a), plasma

lipids, and other lipoproteins with coronary artery disease documented by angiography. Circulation 74:758—765, 1986 7. RHOADs GO, DAHLEN 0, BERG K, MORTON NE, DANNENBERG

AL: Lp(a) lipoprotein as a risk factor for myocardial infarction. J Am Med Assoc 256:2540—2544, 1986 8. Muri A, MIYAHARA T, FUJIMOTO N, MAT5UDA M, KAMEYAMA

M: Lp(a) lipoprotein as a risk factor for coronary heart disease and cerebral infarction. Atherosclerosis 59:199—204, 1986 9. ARMSTRONG VW, CREMER P, EBERLE E, MANKE A, SCHULZE F,

WIELAND H, KREUZER H, SEIDEL D: The association between

serum Lp(a) concentrations and angiographically assessed coronary atherosclerosis, Atherosclerosis 62:249—257, 1986 10. SCHERNTHANER G, KOSTNER GM, DIEPLINGER H, PRAGER R, MUHLHAUSER I: Apolipoproteins (A-I, A-Il, B), Lp(a) lipoprotein and lecithin: Cholesterol acyltransferase activity in diabetes mellitus. Atherosclerosis 49:277—293, 1983 11. TAKEGOSHI T, HABA T, Hirw J, KIT0H C, SAGA T, YAMAZAKI Y,

MABUCHI H: Alterations of lipoprotein(a) in patients with diabetic

patients undergoing hemodialysis or with diabetes mellitus. An nephropathy. Atherosclerosis 83:99—100, 1990 in vivo kinetic study of this type could not be performed in the 12. PARRA HJ, MEZDOUR H, CACHERA C, DRACON M, TACQUET A, present case for ethical reasons. FRUCHART JC: Lp(a) lipoprotein in patients with chronic renal failure treated by hemodialysis. (abstract) Clin Chem 33:721, 1987 When the frequency distribution pattern was assessed by a 13. MURPHY BG, MCNAMEE P, DULY E, HENRY W, ARCHBOLD P, chi square test with cut off points at 10, 20, 30, 40, and 50 mgldl

of Lp(a) (Table 3), significant distribution differences were observed in patient groups at every level of Lp(a) as compared to the control group, but no significant distribution differences were observed between the patient groups. These results also suggest an increased synthesis of apo(a) despite the distribution of apo(a) phenotypes. Chronic renal failure and diabetes mellitus may be linked to the increased levels of Lp(a). Furthermore, they may be secondary factors that increase serum Lp(a) levels.

However, the mechanism which produces an elevated Lp(a) level and the mechanism which produces high and low Lp(a) levels within the same apo(a) phenotype among different mdividuals are still unknown.

TRINICK T: Increased serum apolipoprotein(a) in patients with chronic renal failure treated with continuous ambulatory peritoneal dialysis. Atherosclerosis 93:53—57, 1992

14. HAFFNER SM, GRUBER KK, ALDRETE 0 JR, MORALES PA, STERN

MP, TUTTLE KR: Increased lipoprotein(a) concentration in chronic renal failure. JAm Soc Nephrol 3:1156—1162, 1992 15. KANDOUSSI A, CACHERA C, PAGNIEZ D, DRACON M, FRUCHART

JC, TACQUET A: Plasma level of lipoprotein Lp(a) is high in predialysis of hemodialysis, but not in CAPD. Kidney mt 42:424— 425, 1992 16. OKURA Y, SAKU K, HIRATA K, ZHANG B, LIU R, OGAHARA S, NAIT0 S, KAJIYAMA 0, ARAKAWA K: Serum lipoprotein(a) levels

in maintenance hemodialysis patients. Nephron (in press)

17. UTERMANN 0, MENZEL HJ, KRAFT HG, DUBA HC, KEMMLER

HG, SEITZ C: Lp(a) glycoprotein phenotypes. Inheritance and relation to Lp(a) lipoprotein concentrations in plasma. J Clin Invest

Acknowledgments

80:458—465, 1987 18. UTERMANN 0, KRAFT HG, MENZEL HJ, HOPFERWIEISER T, SEITZ

Part of this work was presented at the 2nd International Conference on Lipoproteina]; New Orleans, Louisiana, USA on November 12—15, 1992. This work was supported by grants-in-aid from the Ministry of Education, Science and Culture of Japan (No. 02671114, No. 04671503,

C: Genetics of the quantitative Lp(a) lipoprotein trait. I. Relation of Lp(a) glycoprotein phenotypes to Lp(a) lipoprotein concentration in plasma. Hum Genet 78:41—46, 1988 19. BOERWINKLE E, LEFFERT CC, LIN J, LACKNER C, CHIESA 0, HOBBS HH: Apolipoprotein(a) gene accounts for greater than 90%

No. 03671090, No. 03265203), and by the research grants from the Ministry of Health and Welfare. Reprint requests to Dr. Kejjiro Saku, Department of Internal Medicine, Fukuoka University School of Medicine, Fukuoka 814-01, Japan.

References

of the variation in plasma lipoprotein(a) concentrations. J Clin Invest 90:52—60, 1992 20. GAUBATZ JW, GHANEM KI, GUEVARA JJ, NAvA ML, PATSCH W,

MORRISETT JD: Polymorphic forms of human apolipoprotein[a}: Inheritance and relationship of their molecular weights to plasma levels of lipoprotein[a]. J Lipid Res 31:603—613, 1990 21. KAMBOH MI, FERRELL RE, KOTTKE BA: Expressed hypervariable polymorphism of apolipoprotein(a). Am J Hum Genet 49:1063— 1074, 1991

1. FLESS GM, ZUMMALLEN ME, SCANU AM: Isolation of apolipoprotein(a) from lipoprotein(a). J Lipid Res 26:1224—1229, 1985 2. FLESS GM, ZUMMALLEN ME, SCANU AM: Physicochemical prop-

22. KIKUCHI S, NAKAGAWA A, KOBAYASHI K, Li L, YANAGi H,

erties of apolipoprotein(a) and lipoprotein(a-) derived from the

ated with plasma lipoprotein(a) levels in Japanese and Chinese

dissociation of human plasma lipoprotein(a). JBiol Chem 261:8712— 8718, 1986

ARINAMI T, KOZU Y, MIYAZAKI R, TSUCHIYA 5, HAMAGUCHI H:

High degree of genetic polymorphism in apolipoprotein(a) associ-

populations. Hum Genet (in press) 23. SAKU K, SASAKI J, NAiT0 S, ARAKAWA K: Lipoprotein and

Hirata et a!: Apolipoprotein(a) phenotypes in hemodialysis

1070

apolipoprotein losses during continuous ambulatory peritoneal dialysis. Nephron 51:220—224, 1989 24. SAKU K, SATA T, NArro S. FUKUSHIMA K, TAKEBAYASHI S,

NOMA A, KAWADE M: Enzyme-linked immunosorbent assay of lipoprotein(a) in serum and cord blood. Gun Chim Acta 177:31—40, 1988

ARAKAWA K: Apolipoproteins in human biopsied nephrotic kidneys. mt Urol Nephrol 20:429—438, 1988 25. NORBECK H-E, ORo L, CARLSON LA: Serum lipid and lipoprotein concentrations in chronic uremia. Ada Med Scand 200:487—492,

35. MILES DW, MOGENSEN CE, GUNDERSEN BiG: Radioimmunoas-

1976 26. GHERARDI E, ROTA E, CALANDRA S, GENOVA R, TAMBORINO A:

LU, OLSSON AG: Significant association between low-molecularweight apolipoprotein(a) isoforms and intermittent claudication,

Relationship among the concentrations of serum lipoproteins and changes in their chemical composition in patients with untreated nephrotic syndrome. Eur J Gun Invest 7:563—570, 1977

Arterioscier Thromb 12:895—901, 1992 37. HAFFNER SM, TUTTLE KR, RAINWATER DL: Lack of change of

27. ATrMAN P0, NYBERG G, WILLIAM-OLSS0N T, KNIGHT-GIBSON C,

ALAUP0vIC P: Dyslipoproteinemia in diabetic renal failure. Kidney

mt 42:1381—1389, 1992 28. ALLAIN CC, PooN LS, CHAN CS: Enzymatic determination of total cholesterol. Clin Chem 20:470—475, 1974 29. EGGESTEIN M, KREUNTZ FH: Em neue Bestimmung der Neutralfette in Biutserum und Gewebe, I. Mitt. Prinzip, Durchfuhrung und Besprechung der Method. (abstract) Kim Wochenschr 44:262, 1966 30. N0MA A, NEZU-NAKAYAMA K, KITA M, OKABE H: Simultaneous

determination of serum cholesterol in high- and low-density iipoproteins with use of heparin Ca2, and anion-exchange resin. Gun Chem 24:1504—1508, 1978 31. FRIEDEWALD WT, LEVY RI, FREDRICKSON DS: Estimation of the

concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem 18:499— 502, 1972 32. HAVEL RL, EDER HA, BRAGTON JH: The distribution and chemical composition of ultracentrifugally separated lipoproteins in human serum. J Gun Invest 34:1345—1353, 1955 33. IKEDA T, SHIBUYA U, SuolucH! H, ARAKI S, UJi Y, OKABE H: Automated immunoturbidimetric analysis of six plasma apolipoproteins: Correlation with radial immunodiffusion assays. J Clin Lab Anal 5:90—95,

1991

34. ABE A, MAEDA S, MAKINO K, SEISHIMA M, SHIMOKAWA K,

say for urinary albumin using a single antibody. Scand J

Clin Lab

1970 36. MOLOAARD J, KLAUSEN IC, LASSVIK C, FAEROEMAN 0, GERDES Invest 26:5—11,

lipoprotein(a) concentration with improved glycemic control in subjects with type II diabetes. Metabolism 41:116—120, 1992 38. HAFFNER SM, Moss SE, KLEIN BEK, KLEIN R: Lack of associ-

ation between lipoprotein(a) concentrations and coronary heart disease mortality in diabetes: The Wisconsin epidemiologic study of diabetic retinopathy. Metabolism 41:194-197, 1992 39. HAFFNER SM, MORALES PA, STERN MP, GRUBER MK: Lp(a) concentrations in NIDDM. Diabetes 41:1267—1272, 1992 40. KREMPLER F, KOSTNER GM, ROSCHER A, HALSLAUER F, BOLZANO K, SANDHOFEK F: Studies on the role of specific cell surface receptors in the removal of lipoprotein(a) in man. J Gun Invest 71:1431—1441,

1983

41. KNIGHT BL, PEROMBELON YFN, SOUTAR AK, WADE DP, SEED

M: Catabolism of lipoprotein(a) in familial hypercholesterolemic subjects. Atherosclerosis 87:227—237, 1991 42. PORTMAN Ri, Scorr RC, ROGERS DD, LOOSE-MITCHELL DS, LEMIRE JM, WEINBERG RB: Decreased low-density lipoprotein receptor function and mRNA levels in lymphocytes from uremic patients. Kidney Int 42:1238—1246, 1992

43. HAVEKES L, VERMEER BJ, BRUGMAN T, EMEIS J: Binding of Lp(a)

to the low density lipoprotein receptor of human fibroblasts. FEBS Len 132:169—173, 1981

44. WINOCOUR PH, HARLAND JOE, MILLAR JP, LAKER MF, ALBERT!

KGMM: Microalbuminuria and associated cardiovascular risk factors in the community. Atherosclerosis 93:71—81, 1992