The relationship between serum levels of lipoprotein(a) and proteins associated with the acute phase response

ELSEVIER

Clinica Chimica Acta 223 (1993) 73-82

The relationship between serum levels of lipoprotein(a) and proteins associated with the acute phase response Thomas B. Ledue*, Louis M. Neveux, Glenn E. Palomaki, Robert F. Ritchie, Wendy Y. Craig

(Received 11 June 1993; revision received 16 August 1993; accepted 17 August 1993)

Abstract The association of serum lipoprotein(a) (Lp(a)) with inflammation was investigated in a primarily rheumatologic study group (n = 570; 202 males and 368 females) by studying the relationship between serum levels of Lp(a) and a panel of acute phase proteins (C-reactive protein (CRP), cx,-antitrypsin (AAT), q-acid glycoprotein (AGP), haptoglobin (HPT), complement components 3 and 4 (C3, C4), prealbumin (PAL), albumin (ALB) and transferrin (TRF)). Lp(a) data were adjusted for age and sex, but not clinical condition as no significant differences in Lp(a) levels were observed, using analysis of variance, among the I5 diagnostic categories in the study group. Univariate analyses revealed significant positive associations between Lp(a) levels and levels of C4, AGP, C3 and HPT. Multivariate analysis revealed that C4 and AGP (in descending order of significance) were significant independent predictors of Lp(a) concentration, together amounting for 2.9% of the variability in Lpfa) concentration in the present study group. The data indicate that confounding effects of an acute phase response should be considered in epidemiologic studies, if a high prevalence of inflammation is suspected. rCey words: Lipoprotein(a);

Acute phase proteins; ~nfla~ation

1. induction

Initially discovered by Berg in 1963 [l], lipoprotein(a) (Lp(a)) has received widespread interest as an independent risk factor for atherosclerotic disease [2]. Its structure is similar to low-density lipoprotein (LDL), in that both contain apolipoprotein B-100 (apo B) and have a similar lipid composition; however, unlike 000%8981/93M6.00 0 1993 Elsevier Science Publishers SSDI 0009-898~(93)057~-K

B.V. All rights reserved.

LDL, Lp(a) contains a highly glycosylated apolipoprotein(a) (ape(a)) which is covalently attached to apo B by disulfide bonds. The structure of ape(a) is strikingly homologous to plasminogen, leading to the suggestion that it may inhibit librinolysis [3]. Recent studies suggest that serum Lp(a) concentration is under polygenic control 141. In Caucasians, serum Lp(a) levels are highly skewed and range from near 0 mgidl to over 100 mg/dl with a median level of approx. 10 mg/dl. Epidemiologic studies have shown that concentrations over 30 m&d1 (- 80th percentile) are associated with a 2- to 3-fold increased risk for coronary heart disease (CHD) [5]. Furthermore, levels of Lp(a) have been reported to be independent of sex, age, diet and other lipid and lipoprotein risk factors and to be refractory to most lipid-lowering drugs [6]. Altered Lp(a) levels are also seen in various clinical conditions other than CHD. Hepatobiliary disease is associated with a 2-fold decrease in Lp(a) [7], whereas diabetes [8], intermittent claudication [9], cancer [lo], certain renal disorders [l l] and pregnancy [12] are associated with increased levels. The acute phase response (APR) includes local and systemic reactions to inflammation or tissue necrosis accompanied by fever, leucocytosis, an elevated erythrocyte sedimentation rate, as well as characteristic changes in the levels of the acute phase proteins. This phenomenon is also known to result in decreased total cholesterol, high density lipoprotein cholesterol and low density lipoprotein cholesterol [13-151 levels. However, while some studies suggest a relationship between Lp(a) and the APR [ 16,171, others do not [18]. Therefore, the relationship between Lp(a) and the APR has not been fully defined. To investigate further the relationship between serum levels of Lp(a) and the acute phase proteins, we have performed a cross-sectional study of these analyses in a clinical population where a high percentage of subjects had active inflammatory processes. 2. Materials and methods 2.1. Subjects The study group consisted of 570 serum samples drawn from adults (age range 19-100 years) that were submitted to our laboratory for protein studies. The majority of these samples (202 males and 368 females) are from Maine residents, of whom 98% are Caucasian (based on the 1990 census). All assays were performed on serum that had been stored at 24°C for no longer than 3 days. Information on diagnosis was available for 91% (n = 518) of the subjects; using this information, subjects were coded either according to major disease categories (rheumatologic, cancer, stroke, infection, diabetes) or according to organ involvement (pulmonary, hematologic, immunuologic, renal, hepatic, or cardiovascular disease) as recorded by the referring physician. Approximately 15% of the study group did not fall into any of these major categories and were therefore grouped as miscellaneous; likewise, those with no diagnosis were coded as missing. 2.2. Protein measurements Levels of prealbumin (PAL), albumin

(ALB), cri-antitrypsin

(AAT),

haptoglobin

T. B. Ledue ef al. I C/in. Chim. Acta 223 11993) 7342

75

(HPT), q-acid glycoprotein (AGP), aY2-macroglobulin(A2M), transferrin (TRF),

complement components 3 and 4 (C3,C4) wereall measuredby immunoturbidimetry on a Roche COBAS FARA (Belleville, NJ) with monospecific antisera, calibrators and methods previously described [19,20]. C-reactive protein (CRP) was analyzed by latex-enhanced nephelometry on a Behring nephelometer (BNA, Behring Diagnostics Inc., Somerville, NJ), as described elsewhere [21]. Purified human apo B was obtained from Biodesign (Kennebunkport, ME). Goat antiserum to human plasminogen (PSM), obtained from INCSTAR (Stillwater, MN) was used to quantify human PSM immunoturbidimetrically according to the method of INCSTAR (Stillwater, MN) using a human serum pool standard (Behring Diagnostics INC., Somerville, NJ) as a calibrator. The delipidated serum reference pool for the evaluation of Lp(a) assay specificity (Apo A-I < 2 mgidl; apo B < 0.5 mg/dl; Lp(a) undetectable; cholesterol < 1 mg/dl and triglyceride < 2 mg/dl) was obtained from Behring Diagnostics Inc. (Marburg, Germany). 2.3. Measurement of Lp(a) Lp(a) was measured with an enzyme-linked immunosorbent assay (ELISA) from Terumo (Elkton, MD). To determine whether the Lp(a) assay was confounded by interference from either human apo B or PSM, a purified human apo B preparation was diluted in delipidated human serum to levels corresponding to normal, 2- and 4-fold normal human serum concentrations and assayed for Lp(a). Similarly, a delipidated serum (PSM, 24 mg/dl) was added to the ELISA plate at PSM concentrations from normal to 4-fold normal human serum concentrations. The results, illustrated in Table 1, demonstrate that there was minimal (< 1%) cross-reactivity in the ELISA assay with either human apo B or PSM. 2.4. Statistical methods When parametric statistical methods were necessary, serum protein that were non-Gaussian were transformed by taking the logarithm or Differences between subgroups were analyzed by the Kruskal-Wallis analyte. Differences in Lp(a) levels among diagnostic categories in the were analyzed by ANOVA after log transformation.

Table 1 Cross-reactivity

in the Lp(a) ELISA

distributions square root. test for each study group

assay

Human Apo B (lot 29-B2227a) (mg/dl)

Delipidated human serum pool as a source of plasminogen (mgdl)

Apo B added

Lp(a) detected

PSM added

Lp(a) detected

0 88 177 354


25 50 75 100


limitb

aApo B purity > 98% by SDS-PAGE according to manufacturer. bLower limit of detection = 0.15 mg/dl (2 S.D. above the mean absorbance

limitb

for the zero calibrator).

The relationship between Lp(a) and age was investigated by linear regression among individuals without an APR, as defined by CRP levels ~0.85 mg/dl. This cutoff is consistent with the upper limit of CRP in normal healthy adults [22]. To adjust Lp(a) levels for age and sex, separate median Lp(a) levels were found for males and females for each 10 year increment of age. Using a polynomial regression model for males and females separately, smoothed Lp(a) levels for a given age and sex were then estimated. An individual’s Lp(a) level was then expressed as a multiple of this age and sex-specific median value [23]. In order to adjust the rest of the data for age and sex, the other serum protein concentrations were also expressed as multiples of the median (MOMS); a separate (larger) reference database was used to obtain age- and sex-specific medians for this purpose. This reference database also comprised data from samples received for clinical testing and was a subset of our total clinical database from 1979 to 1984. All cases with a referral diagnosis involving a major organ system (e.g. heart, liver, kidney, lungs, endocrine) or a major disease category (e.g. inflammation, infection, cancer, arthritis, atherosclerosis, autoimmune disorder) were excluded. Furthermore, the reference database was trimmed to exclude any case with concentrations of one or more analyte that were > 3 S.D. from the mean value for subjects in the same age and sex category. The final reference database included 18,420 males and 32,673 females, age newborn to 100 years. Separate median protein levels were found for males and females for each 10 year age increment and serum protein MOMS in the present study group were calculated as described above for Lp(a). The contribution of the other measured variables to the variability in Lp(a) MOM levels was determined by stepwise linear regression. Linear regression was used to test for trend in Lp(a) MOM level among the quintiles of each serum protein. Statistical significance was accepted at P < 0.05. All analyses were performed with a statistical package from BMDP Statistical Software Inc. (Los Angeles, CA). 3. Results 3.1. The effects of potential confounding factors on serum Lp(a) concentration We studied the effects of age and sex on Lp(a) levels among subjects without evidence of an APR as indicated by normal CRP levels (see Methods). This approach for selecting individuals without an APR was validated by comparing serum protein levels between the two subgroups (Table 2). We observed no significant difference in Lp(a) levels (mean, median) between males (18.5, 8.8 mg/dl, n = 158) and females (18.6, 10.6 mg/dl, n = 289); however, the relationship between Lp(a) and age differed slightly between males and females. Lp(a) levels increased similarly with age in males (log Lp(a) = 0.741 + 0.0032 x age, r = 0.089) and females (log Lp(a) = 0.714 + 0.0052 x age, r = 0.149). The effect of age on Lp(a) levels although similar was only significant among females (P = 0.01, P = 0.3 for females and males, respectively). Therefore, Lp(a) data were adjusted for age and sex as described in Methods. To examine the effects of clinical condition on Lp(a) within the study group, study subjects were grouped into 15 diagnostic categories, as described in Methods. There

77 Table

2

Descriptive statistics forserum proteins intheacute phase response (AIR)subgoups asstratified $ C-reactive

protein APR(-) (n = 447)

C-reactive

protein

0.34 (0.36)b 0.2

Prealbumin Albumin Transferrin AAT AGP Haptoglobin Complement

component

3

Complement

component

4

A2M

P”

APRf+) (n = 123) 5.47 (9.04) 2.2

31.0 (6.0) 31 4104 (372) 4130

24.5 (7.X) 24 3477 (522) 3580

<0.0001

285 (40.1) 280 214 (37.4) 220

249 (54.3) 260 299 (89.0) 280

<0.0001

68.5 (16.X) 66 I45 (58.0) I41

<0.0001

I I8 (23.7) I16 28.4 (9.0) 28

I I8 (38.0) IOU 257 (105) 260 I45 (31.5) 142 33. I (10.4) 32

173 (47.6) I60

166 (42.8) 160

0.1

“Differences between subgroups were analyzed bValues are mgfdl (mean. (SD); median).

by the Kruskal-Wallis


<0.0001

<0.0001 <0.0001 0.0003

test.

were no si~i~~ant differences in Lp(a) levels among these categories (Table 3); therefore, Lp(a) was not adjusted for clinical condition in subsequent analyses. 3.2. The relationship between serum levels of Lp(a) and the acute phase proteins The relationships between Lp(a) MOM levels and each of the different serum acute phase proteins tested are displayed in Table 4. Concentrations of Lp(a) and other assayed proteins were adjusted for age and sex, as described in Methods. Lp(a) levels demonstrated significant positive trends with AGP, HPT, C3 and C4. Multiple stepwise regression was used to identify a combination of acute phase proteins that were most predictive of variation in Lp(a). Of the four acute phase proteins determined to be univariately associated with Lp(a), only C4 and AGP (in decreasing order of significance) showed statistically significant association by multiple stepwise regression. The incremental changes in Y’ were 0.023 for C4 and 0.006 for AGP. When combined, the two variables accounted for 2.9% of the variation in Lp(a) MOM levels within the present study group.

78

Table 3 Clinical category

T. B. Lrdue er ul. / Clin. Chim. Aciu 223 (1993)

73-82

and Lp(a) concentration

Category

n

Diabetes mellitus Immunologic disease Neural/eyes/ears Cardiovascular disease Hematologic disease Hypertension Infection Renal disease Pulmonary disease

3 25 52 3 83 256 12 38 23 26 16 7 IX 3 s

Total

570

Hepatic disease Cancer Not available Stroke Miscellaneous Rheumatologic disease

‘%Iof total

0.5

4.4 9. I 0.5 14.6 44.9 2.1 6.7 4.0 4.6 2.x I.’ 3.2 0.5 0.9

Median MOM“

0.24 0.75 0.77 0.82 0.94

1.20 1.24 I.25 1.2x I .43 1.47 1.56 1.63 1.x’) 2.05

Lp(a)

Geometric

mean

Lp(a) MOM. (log S.D.)h 0.12 (NC) 0.82 (0.58) 0.59 (0.63) 0.96 (NC) 0.78 (0.72) I .OO (0.60) 0.95 (0.66) 0.87 (0.64) 0.87 (0.75) I .06 (0.62) 0.91 (0.55) 2.03 (NC) 1.03 (0.58) I .05 (NC) 1.23 (NC)

100

dMultiple of the Median (MOM) adjusted for age and sex hNC. S.D. not calculated if n < IO. Levene’s test for variance gave F = 0.5 I and P = 0.9. Welch’s equality of means test gave F= 0.75 and P = 0.6.

infarction or surgical operation, Lp(a) levels increased transiently, in concert with the levels of CRP, AGP, AAT and HPT. Similarly, Dahlen [ 171 noted significantly higher mean Lp(a) levels in serum from 50 males, aged 60-63 years, with elevated acute phase proteins, compared with controls. In contrast, Slunga et al. [ 181 found no relationship between Lp(a) and AGP, HPT, or AAT following acute myocardial infarction and suggested that the Lp(a) increases seen after myocardial infarction are linked to the reaction of the lipoproteins, rather than to the acute phase proteins. Conversely, Oshima et al. [24] report that Lp(a) levels increase transiently in unstable angina, but without any corresponding changes in CRP or AAT levels. In each of these reports, the size of the study groups were small (32 to 36), thus contradictory results are not unexpected. To investigate this question further, we examined the relationship between serum Lp(a) levels and levels of a comprehensive panel of acute phase proteins in a study group where the incidence of inflammation is high. Indeed, based on CRP data, 22% of the study group demonstrated an active acute phase response. Serum protein lindings were consistent with this designation. The current study found that Lp(a) levels increased with age in both males and

7.1.ledtie et al.JGIL. Ch.

Acta 22311993) 7%b2

^_____1__-

*va~~mr-‘noo-r: 9 01’: --00-000--d l-_-______

9

Q: 0:

01?

CJt

ding that Lp(a) levels increase with age contrasts with data from earlier studies, where no association was found between Lp(a) and age [2,26,27]; indeed, it has been widely assumed that, after the first 2 years, Lp(a) levels do not change with age [28]; however, there have been reports, consistent with the present findings, that Lp(a) levels do in fact increase with age [25,28,30]. Altered Lp(a) levels are known to be associated with several of the included diagnostic categories [5-l 11; however, the number of such cases in the study group was too small to be confounding, as evidenced by the non-significant result of ANOVA. In the present study, we have been able to demonstrate a significant association between serum levels of Lp(a) and the levels of certain serum proteins (C4, AGP, C3 and HPT). In contrast, there was no significant trend in Lp(a) levels with increasing concentrations of A2M, a protein that is believed to be minimally influenced by the presence of an acute phase response [3 11. Furthermore, in a multivariate analysis, C4 and AGP were relatively independent predictors of Lp(a) MOM levels and together accounted for 2.9% of the variability. The AGP data are consistent with the findings of Maeda et al. [16], who reported that the longitudinal changes in Lp(a) and AGP were similar following acute myocardial infarction or surgery. This may in part explain why CRP levels did not show a significant association with Lp(a). A similar finding, that Lp(a) levels were significantly associated with levels of AGP and HPT but not CRP, has been recently reported in a population of patients with rheumatoid arthritis [32]. The relationship between C4 and Lp(a) has not been documented previously, although a relationship between C4 and other lipoproteins has been reported [33]. Even though the impact of altered acute phase protein levels on serum Lp(a) levels is small, it is statistically significant; thus an acute phase response could be taken into account when designing or interpreting epidemiologic studies, especially if a high incidence of inflammation is suspected. This approach has already been acknowledged by some investigators; Fieseler et al. [34] used CRP levels to identify cases potentially confounded by an acute phase response. However, the present multivariate data indicate that C4 and AGP are the most predictive analyses in determining whether Lp(a) data are confounded by the presence of an acute phase response. 5. Acknowledgement This study was funded by the Foundation

for Blood Research

Development

Fund.

6. References 1 2 3 4 5 6

Berg K. A new serum type in man - the Lp system. Acta Pathol Microbial Stand 1963;59:369-389. Sandkamp M, Funke H, Schulte J, Kohler E, Assman G. Lipoprotein(a) is an independent risk factor for myocardial infarction at a young age. Clin Chem 1990;36:20-23. Loscalzo J, Weinfeld M, Fless GM, Scanu AM. Lipoprotein(a), fibrin binding, and plasminogen activation. Arteriosclerosis 1990;10:240-245. Hasstedt SJ, Williams RR. Three alleles for quantitative Lp(a). Genet Epidemiol 1986;3:53-55. Kostner, GM, Avogaro P, Cazzolato G, Marth E, Bittolo-Bon H, Quinci GM. Lipoprotein(a) and the risk for myocardial infarction. Atherosclerosis 1981;38:51-61. Houlston R, Fried1 W. Biochemistry and clinical significance of lipoprotein(a) (Review). Ann Clin Biochem 1988:25:499-503.

T. B. Ledue et al. / Clin. Chim. Arta 223 (19931 7342

XI

7 YamashiroM, SeishimaM, KawadeM. Alterationsin serumLp(a) lipoproteinconcentrationsin

patients with hepatobiliary disorders, Jpn JClin Chem 1987;12:79, 8 9

10 11 12 13 14 15 16 17

18 19 20 21 22 23

24

25

26 27

28 29 30 31

Bruckert E, Davidoff P, Grimaldi A et al. Increased serum levels of lipoprotein(a) in diabetes mellitus and their reduction with glycemic control. J Am Med Assoc 1990;263:35-36. Molgaard J, Klausen IC, Lassvik C, Faergeman 0, Gerdes LU, Olsson AG. Significant association between low-molecular-apolipoprotein(a) isoforms and intermittent claudication. Arthritis Thromb 1992;12:895-901. Wright L, Sullivan D, Muller M, Dyne M. Tattersall, M, Mountfort C. Elevated apolipoprotein(a) levels in cancer patients. Int J Cancer 1989;43:241-244. Karadi J, Romics L, Palos G et al. Lp(a) lipoprotein concentration in serum of patients with heavy proteinuria of different origin, Clin Chem 1989;35:2121-2123. Zechner R, Desoye G, Schweditsch M, Pfeiffer K, Kostner G. Fluctuations of plasma lipoprotein(a) concentrations during pregnancy and postpartum. Metabolism 1986;35:333-336. Coombes EJ, Shakespeare PG, Batstone GF. Lipoprotein changes after burn injury in man. J Trauma 1980;20:971-975. Alvarez C, Romos A. Lipids, lipoproteins and apolipoproteins in serum during infection. Clin Chem 1986;32:142-145. Boille EE, Orr CW. Lowered high-density lipoprotein cholesterol in viral illness. Clin Chem 1979;25:817-818. Maeda S, Abe A, Seishima M, Makino K, Noma A. Kawade, M. Transient changes of serum lipoprotein(a) as an acute phase protein. Atherosclerosis 1989;78: 144- 150. Dahlen GH. Lipoprotein(a) in relation to athersclerotic diseases, In: K. Widhalm, HK Naito, eds. Recent aspects of diagnosis and treatment of lipoprotein disorders: impact on prevention of atherosclerotic diseases. New York: A.R. Liss, 1988;27-36. Slunga L, Johnson 0, Dahlen GH, Eriksson S. Lipoprotein(a) and acute phase proteins in acute myocardial infarction. Stand J Clin Lab Invest 1992;52:95-101. Hudson G, Poulin S, Ritchie R. Twelve-protein immunoassay profile on the COBAS Lab Anal 1987;1:191-197. Ledue TB, Rifai N, Irish GR, Silverman LM. Immunoturbidimetry of transthyretin in human serum. Clin Chem 1987;33:1260.

FARA.

J Clin

(prealbumin)

Kapmeyer W, Pauly JE, Tuengler P. Automated nephelometric immunoassays with novel shell/core particles. J Clin Lab Anal 1988;2:76-83. Pepys MB. C-reactive protein fifteen years on. Lancet 1981;653-657. Wald J, Cuckle HS. Maternal serum alpha-fetoprotein measurement in antenatal screening for anencephaly and spina bifrda in early pregnancy: report of the UK collaborative study on alphafetoprotein in relation to neural-tube defects. Lancet 1977; 1:1323- 1332. Oshima S, Uchida K, Yasu T, Uno K, Ninogu H, Haze K. Transient increase of plasma lipoprotein(a) in patients with unstable angina pectoris. Does lipoprotein(a) alter Iibrinolysis? Arterioscler Thromb 1991;11:1772-1777. Jenner JL, Ordovas JM, Lamon-Fava S, Schaefer MM, Wilson PWF, Castelli WP, Schaefer EJ. Effects of age, sex and menopausal status on plasma lipoprotein(a) levels: the Framingham offspring study. Circulation 1993;4: 1135- 1141. Albers JJ, Hazzard WR. Immunochemical quantification of human plasma Lp(a) lipoprotein. Lipids 1974;9: 15-26. Seed M, Hoppichler F, Reaveley D, McCarthy S, Thompson GR, Boerwinkle, E, Utennann G. Relation of serum lipoprotein(a) concentration and apolipoprotein(a) phenotyope to coronary heart disease in patients with familial hypercholesterolemia. N Engl J Med 1990;322:1494-1499. Rifai N, Heiss G, Doetsch K. Lipoprotein(a) at birth, in blacks and whites. Atherosclerosis 1992;92:123-129. Alexander RG, Seboldt RP. The quantification of Lp(a) in serum and its application as a marker of coronary atherosclerosis. Clin Chem 1990;36:961. Guyton JR, Dahlen GH, Patsch W, Kautz JA, Gotto AM. Relationship of plasma lipoprotein Lp(a) levels to race and to apolipoprotein B. Arteriosclerosis 1985;5:265-272. Crockson RA, Payne CJ, Ratcliff AP, Soothill JF. Time sequence of acute phase reactive proteins following surgical trauma. Clin Chim Acta 1966; 14:435-441.

82

32 33 34

T. B. Ledue 6’1cd. / Clin. C&m

Ac,ru 223 i 1993) 7342

Rantapaa-Dahlqvist S, Wallberg-Johnson S, Dahlen G. Lipoprotein(a), lipids, and lipoproteins in patients with rheumatoid arthritis. Ann Rheum Dis 1991;50:366-368. Muscari A, Bozzoli C, Gerratama C et al. Association of serum IgA and C4 with severe atherosclerosis. Atherosclerosis 1988;74: 179-186. Fieseler HG, Armstrong VW, Wieland E et al. Serum Lp(a) concentrations are unaffected by treatment with the HMG-CoA reductase inhibitor pravastatin: results of a 2-year investigation, Clin Chim Acta 1991;204:291-300.