- Email: [email protected]
Apolipoprotein E predicts incident cardiovascular disease risk in women but not in men with concurrently high levels of high-density lipoprotein cholesterol and C-reactive protein
M ET ABOL I S M CL IN I CA L A N D EX PE R IM EN TA L 6 1 ( 2 0 12 ) 99 6 –1 0 02
Available online at www.sciencedirect.com
Metabolism www.metabolismjournal.com
Apolipoprotein E predicts incident cardiovascular disease risk in women but not in men with concurrently high levels of high-density lipoprotein cholesterol and C-reactive protein James P. Corsetti a,⁎, Ron T. Gansevoort b , Stephan J.L. Bakker b , GerJan Navis b , Charles E. Sparks a , Robin P.F. Dullaart c a b c
Department of Pathology and Laboratory Medicine, University of Rochester School of Medicine and Dentistry, Rochester, NY 14642, USA Department of Nephrology, University of Groningen and University Medical Center Groningen, 9700 RB Groningen, the Netherlands Department of Endocrinology, University of Groningen and University Medical Center Groningen, 9700 RB Groningen, the Netherlands
A R T I C LE I N FO
AB S T R A C T
Article history:
Although there is great interest in the notion that dysfunctional transformation of high-
Received 13 October 2011
density lipoprotein (HDL) facilitates development of atherosclerosis and cardiovascular
Accepted 19 November 2011
disease (CVD), few studies in human populations directly address this issue. As apolipoprotein E (apoE) is a constituent of HDL thought to be important for HDL antiatherogenic function, we sought to assess the role of apoE in CVD risk in subjects likely to display dysfunctional transformation of HDL. Association of apoE levels with incident CVD risk was investigated using Cox multivariable proportional hazards modeling. Analyses were performed in subgroups of women and men likely to display dysfunctional transformation of HDL deriving from previous subgroup identification based upon defining characteristics of concurrently high levels of HDL cholesterol and systemic inflammation as reflected by high C-reactive protein levels. Results revealed apoE levels (dichotomized as highest quartile vs combined 3 lowest quartiles) as predicting subgroup risk in women (hazard ratio, 4.52; 95% confidence interval, 1.07-19.12; P = .040) but not in men. Further sex differences were manifested in terms of the relationship of apoE levels with age. Analysis revealed positive correlation of apoE levels with age in women (r = 0.47, P < .0001) but not in men (r = 0.04, P = .43). Apolipoprotein E levels predict incident CVD risk in women with high levels of HDL cholesterol and C-reactive protein but not in men. Future studies should be oriented toward investigations of apoE as related to multiplicity of HDL functionality and toward assessment of potential roles for apoE in dysfunctional transformation of HDL. © 2012 Elsevier Inc. All rights reserved.
1.
Introduction
In addition to the traditional focus on high-density lipoprotein (HDL) levels, there is growing interest in the role of HDL
functionality in relation to atherogenesis and cardiovascular disease (CVD). In this context, HDL antiatherogenic functionality is thought to be based upon multiple protective roles including cholesterol clearance by reverse cholesterol transport (RCT),
Author contributions: JPC: study conception and design, data analysis and interpretation, manuscript drafting; RTG: data acquisition, manuscript editing; SJLB: data acquisition, manuscript editing; GJN: data acquisition, manuscript editing; CES: study conception and design, data analysis and interpretation, manuscript editing; RPFD: study conception and design, data analysis and interpretation, manuscript editing. ⁎ Corresponding author. Department of Pathology and Laboratory Medicine, University of Rochester Medical Center, Box 626, Rochester, NY 14642, USA. Tel.: +1 585 275 4907; fax: +1 585 273 3003. E-mail address: [email protected] (J.P. Corsetti). 0026-0495/$ – see front matter © 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.metabol.2011.11.010
M ET ABOL I S M CL IN I CA L A N D EX PE R IM EN TA L 6 1 ( 2 0 1 2) 9 96 –1 0 02
resistance to vascular inflammatory and oxidative stress processes, preservation of endothelial function, and antithrombotic effects [1-3]. It is now widely recognized that HDL particles can become dysfunctional in the setting of inflammation and oxidative stress [4-8]. Such considerations are especially important in view of intense efforts currently under way to reduce residual CVD risk by raising HDL cholesterol (HDL-C) levels. However, despite potential implications of dysfunctional transformation of HDL in regard to CVD risk, little work is available addressing this issue in human populations. To this end, we have studied associations of CVD risk in humans with high HDL-C levels in the setting of inflammation as reflected by high C-reactive protein (CRP) levels. Subjects with high HDL-C levels were chosen for the study, as risk associated with dysfunctional HDL transformation would be difficult to discern in subjects with low HDL-C levels given the wellknown inverse correlation with risk. In addition, it should be noted that high levels of HDL-C have been associated with risk under various circumstances as reported in earlier studies [920]. Thus, for subjects with relatively high levels of both HDL-C and CRP, we have identified high risk in men and women without previous history of CVD in the Prevention of Renal and Vascular End-Stage Disease (PREVEND) study group [21,22] as well as in postinfarction patients in the Thrombogenic Factors and Recurrent Coronary Events Postinfarction study group [23]. Moreover, these studies used functional genetic polymorphisms of relevant genes to provide evidence in support of involvement in risk of oxidative stress (CYBA) [24] and early phases of RCT (CETP, LPL) [22,23]. Antiatherogenic properties of HDL are thought to derive from multiple HDL particle constituents including apolipoproteins (apoA1, apoA2, and apoE) and enzymes (paraoxonase 1, lecithin-cholesterol acyltransferase, and glutathione peroxidase) [7]. With particular regard to apoE, antiatherogenic properties are usually ascribed to effects on lipoprotein particle clearance in conjunction with RCT [25,26]. However, in addition to its role in RCT, evidence is accumulating that apoE has significant immunoregulatory properties that could also play a role in apoE-associated antiatherogenic function [27]. In view of the multiple potential protective roles for apoE in terms of atherogenesis and CVD, we assessed effects on incident cardiovascular disease risk of plasma apoE levels in previously identified high-risk subgroups of men and women [21,22] defined by concurrently high levels of HDL-C and CRP from the PREVEND population study [28].
2.
Subjects and methods
2.1.
Study populations
The current work was based on PREVEND [28], a prospective longitudinal study of albuminuria in predicting cardiovascular [29] and renal disease [30]. PREVEND was approved by the medical ethics committee of the University of Groningen, the Netherlands; informed consent was obtained from all subjects. Preliminary study exclusions were for insulin-using diabetic patients and pregnancy. Further exclusions for the current study included history of diabetes mellitus, renal disease, previous CVD, incomplete laboratory results, and CRP
997
levels of at least 10 mg/L (to avoid confounding by intercurrent illness). Outcome events included CVD mortality and CVDrelated hospitalization (acute myocardial infarction, acute and subacute ischemic heart disease, coronary artery bypass grafting, and percutaneous transluminal coronary angioplasty). Median follow-up time was 7.6 years. Detailed descriptions of definitions and data acquisition techniques were reported previously [21,28-31].
2.2.
Blood biomarkers
Serum and plasma biomarkers were analyzed on samples stored at −20°C prepared from venous blood after overnight fast with collection after 15 minutes of rest. Total cholesterol, HDL-C, triglycerides, apoA1, apoB, glucose, high-sensitivity CRP, creatinine, and urine albumin were determined as described previously [21,31]. Urinary albumin excretion rate (UAE) is the mean of two 24-hour collections. Apolipoprotein E levels were determined by immunonephelometry (BNII; Dade Behring, Marburg, Germany). Low-density lipoprotein cholesterol (LDL-C) levels were estimated using the Friedewald equation.
2.3.
Statistical analyses
Analyses were performed using Statistica 10.0 (StatSoft, Tulsa, OK) including Mann-Whitney U, Kruskal-Wallis with Bonferroni correction, and Cox proportional hazards multivariable regression with adjustment for clinical covariates (age, hypertension, body mass index [BMI], ethanol use, and smoking) and urine and blood biomarkers (UAE, creatinine, apoA1, apoB, cholesterol, CRP, HDL-C, triglycerides, LDL-C, and glucose) based on significance (P < .10) in univariate analysis. This approach was taken to minimize inclusion of potentially extraneous predictor variables in multivariable models. Apolipoprotein E was treated as a dichotomized variable (highest quartile vs combined lowest 3 quartiles). Statistical significance was at the P < .05 level. Outcome event mapping [22,24,32], a graphical exploratory data analysis tool, was used to identify high-risk subgroups as a function of 2 risk parameters. In this case, risk parameters were HDL-C and CRP levels. A 3-dimensional scatter plot is generated with CVD outcome on the zaxis (coded 0 for no outcome and +1 for outcome) vs the 2 rank-transformed (to more evenly distribute patients over the bivariate risk domain) risk parameters (x-axis, HDL-C levels; y-axis, CRP levels). Application of a smoothing algorithm results in a surface (outcome event map) generally demonstrating peaks and valleys and where height above the bivariate x-y plane approximates the outcome rate. Subjects contained within peaks in the mappings comprise high-risk subgroups. Peaks are defined as regions with outcome rates greater than the mean outcome rate of the total population. In practice, a contour line representing the constant value of the mean outcome rate is generated on the mapping. In the present case, the mean outcome rate was 1.9% for women and 5.5% for men. High-risk subjects are then identified as those subjects within the footprint (projection) cast on the x-y plane by the contour line corresponding to the value of the mean outcome rate.
998
M ET ABOL I S M CL IN I CA L A N D EX PE R IM EN TA L 6 1 ( 2 0 12 ) 99 6 –1 0 02
Table 1 – Clinical and laboratory characterization (mean ± SD) and comparisonsa among BG, HR1, and HR2 subjects in female participants. BG (n = 2232)
CVD outcomes % (n) Clinical Age (y) BMI (kg/m2) Hypertension (%) Metabolic syndrome (%) Alcohol use (%) Smoking status (%) Never Former Current Biomarkers CRP (mg/L) c
1.03 (23)
3.53 (31)
2.95 (15)
46.2 ± 11.6 24.2 ± 3.5 27.2
50.1 ± 12.3 28.2 ± 5.0 49.0
48.5 ± 12.9 27.0 ± 4.8 45.3
.02 <.001 .18
3.53
46.29
12.35
<.001
20.2
12.5
17.0
.02
HDL-C (mmol/L) ApoE (g/L) ApoA1 (g/L) Triglycerides (mmol/L) ApoB (g/L) Cholesterol (mmol/L) LDL-C (mmol/L) Glucose (mmol/L) Creatinine (μmol/L) UAE (mg/24 h) c
HR1 (n = 879)
HR2 (n = 508)
Pb
Parameter
.57
.047 35.8 34.3 30.0
31.9 28.4 39.7
31.9 34.1 34.1
0.64 (0.34-1.25) 1.67 ± 0.38
2.43 (1.47-4.31) 1.07 ± 0.16
4.66 (3.35-6.42) 1.55 ± 0.21
<.001
0.0365 ± 0.0128 1.54 ± 0.30 0.99 ± 0.46
0.0422 ± 0.0180 1.28 ± 0.20 1.74 ± 1.03
0.0377 ± 0.0133 1.59 ± 0.25 1.27 ± 0.55
<.001
0.91 ± 0.26 5.47 ± 1.11
1.15 ± 0.33 5.87 ± 1.19
1.01 ± 0.27 5.67 ± 1.17
<.001 .0015
3.35 ± 1.04
4.00 ± 1.04
3.53 ± 1.09
<.001
4.50 ± 0.61
4.98 ± 1.32
4.71 ± 0.85
<.001
75.8 ± 11.8
76.2 ± 12.8
76.3 ± 10.1
.44
7.8 (5.6-12.2)
9.4 (6.2-18.4)
8.9 (5.8-15.1)
.036
<.001
<.001 <.001
ed as HR1); and the other, by high levels of HDL-C and high levels of CRP (henceforth designated as HR2). Lower-risk female subjects and lower-risk male subjects constituted background subgroups (henceforth designated BG). Clinical and laboratory characterization for the subgroups is given in Table 1 for women and in Table 2 for men. Briefly, women in HR1 vs HR2 were slightly older; exhibited more features of metabolic syndrome; used alcohol less; and had higher levels of apoE, cholesterol, LDL-C, triglycerides, glucose, and apoB, and lower levels of HDL-C, CRP, and apoA1. Men in HR1 vs HR2
Table 2 – Clinical and laboratory characterization (mean ± SD) and comparisonsa among BG, HR1, and HR2 subjects in male participants. BG (n = 1813)
CVD outcomes % (n) Clinical Age (y) BMI (kg/m2) Hypertension (%) Metabolic syndrome (%) Alcohol use (%) Smoking status (%) Never Former Current Biomarkers CRP (mg/L) c
2.87 (52)
10.08 (77)
7.22 (27)
46.1 ± 12.1 25.0 ± 3.2 47.9
51.6 ± 12.0 28.0 ± 3.8 64.2
54.6 ± 12.9 26.5 ± 3.9 67.2
<.001 <.001 .31
9.65
53.99
7.67
<.001
34.3
29.4
44.5
<.001
HDL-C (mmol/L) ApoE (g/L)
a
Kruskal-Wallis with Bonferroni correction with results as follows: (1) BG vs HR1—all parameters significantly different except for creatinine; and (2) BG vs HR2—all parameters significantly different except for alcohol use, distribution of smokers, apoE, and creatinine. b Comparison of HR1 vs HR2. c For CRP and UAE, median values and interquartile ranges are given.
ApoA1 (g/L) Triglycerides (mmol/L) ApoB (g/L) Cholesterol (mmol/L) LDL-C (mmol/L) Glucose (mmol/L) Creatinine (μmol/L) UAE (mg/24 h) c
The resultant outcome event maps for the current study have been reported previously [21,22].
3.
Results
3.1.
Study populations
a
Previous results using outcome event mapping, a graphical exploratory data analysis tool [23,31], revealed 2 similar highrisk subgroups in the female [22] and male [21] subcohorts. One subgroup was characterized by relatively low levels of HDL-C and relatively high levels of CRP (henceforth designat-
HR1 (n = 764)
HR2 (n = 374)
Pb
Parameter
.12
.004 34.0 40.4 25.6
18.4 32.9 48.7
15.5 43.1 41.4
0.53 (0.30-0.81) 1.28 ± 0.32
2.79 (1.78-4.73) 0.86 ± 0.13
3.60 (2.53-5.22) 1.40 ± 0.25
<.001
0.0369 ± 0.0151 1.34 ± 0.26 1.34 ± 0.93
0.0461 ± 0.0239 1.15 ± 0.17 2.25 ± 1.78
0.0426 ± 0.0162 1.47 ± 0.23 1.32 ± 0.67
.26 <.001 <.001
1.00 ± 0.27 5.50 ± 1.03
1.21 ± 0.31 5.92 ± 1.26
1.11 ± 0.28 5.97 ± 1.13
<.001 .51
3.60 ± 0.93
4.03 ± 1.18
3.96 ± 1.03
.19
4.75 ± 0.77
5.26 ± 1.47
5.05 ± 1.31
.001
90.2 ± 12.3
93.9 ± 37.2
92.7 ± 17.3
.32
9.0 (6.4-15.5) 12.7 (7.5-30.3) 11.9 (7.4-29.2)
<.001
.44
Kruskal-Wallis with Bonferroni correction with results as follows: (1) BG vs HR1—all parameters significantly different; and (2) BG vs HR2—all parameters significantly different except for metabolic syndrome, HDL/apoA1, triglycerides, apoB/apoA1, and creatinine. b Comparison of HR1 vs HR2. c For CRP and UAE, median values and interquartile ranges are given.
999
M ET ABOL I S M CL IN I CA L A N D EX PE R IM EN TA L 6 1 ( 2 0 1 2) 9 96 –1 0 02
Table 3 – Clinical and laboratory characterization (mean ± SD) and comparisonsa among the HR2 women as a function of CVD events. Parameter
Clinical Age (y) BMI (kg/m2) Hypertension (%) Metabolic syndrome (%) Alcohol use (%) Smoking status (%) Never Former Current Biomarkers CRP (mg/L) b HDL-C (mmol/L) ApoE (g/L) ApoA1 (g/L) Triglycerides (mmol/L) ApoB (g/L) Cholesterol (mmol/L) LDL-C (mmol/L) Glucose (mmol/L) Creatinine (μmol/L) UAE (mg/24 h) b a b
HR2 women HR2 women without events with events (n = 493) (n = 15)
P
48.2 ± 12.8 26.9 ± 4.8 44.2 12.3
60.3 ± 9.6 29.8 ± 5.3 80.0 13.3
<.001 .014 .006 .91
16.9
20.0
31.9 34.7 33.5
33.3 13.3 53.3
4.60 (3.34-6.39) 1.55 ± 0.21 0.0374 ± 0.0131 1.59 ± 0.25 1.26 ± 0.55 1.00 ± 0.27 5.63 ± 1.16 3.50 ± 1.08 4.68 ± 0.79 76.2 ± 10.0 8.8 (5.8-14.9)
5.64 (4.46-7.23) .082 1.55 ± 0.16 .69 0.0505 ± 0.0103 <.001 1.58 ± 0.22 .99 1.63 ± 0.52 .003 1.23 ± 0.27 .002 6.81 ± 1.04 <.001 4.51 ± 0.99 <.001 5.37 ± 2.00 .15 77.7 ± 13.3 .88 13.4 (6.9-25.2) .13
.75 .16
Mann-Whitney U test. For CRP and UAE, median values and interquartile ranges are given.
demonstrated similar differences except for younger age and no differences in apoE, cholesterol, and LDL-C.
and LDL-C; for men—older age, more hypertension, and higher urinary albumin excretion. Results of multivariable modeling of potential associations of apoE levels with risk within the BG, HR1, and HR2 subgroups in women and men are given in Table 5. For HR1 in women, there was a loss of significance of apoE levels in the multivariable model relative to the univariate result, whereas for HR2 in women, apoE levels maintained significance as a predictor of risk in the multivariable model. For men, apoE levels did not significantly predict risk in the HR1 and HR2 subgroups. Analyses were carried out using Cox proportional hazards modeling using dichotomized apoE levels first in univariate Cox analysis followed by multivariable modeling adjusted for significant clinical covariates and blood marker levels as described in “Subjects and Methods.” To investigate the lack of apoE-associated risk in HR2 men in contrast to HR2 women, event rates for the 2 subgroups were compared. The event rate for HR2 men was 7.22%; and as noted, it was not a function of apoE levels. The event rates for HR2 women with low vs high apoE levels were 0.88% and 7.02%, respectively. Thus, the event rate in HR2 women with high apoE levels was almost exactly the same as the event rate in HR2 men; however, the event rate in HR2 women with low apoE levels was close to an order of magnitude smaller. To explore whether age might contribute to the sex difference, apoE levels as a function of age in HR2 women were investigated. Mean age for women with low apoE levels (lowest 3 quartiles) was 45.5 ± 12.1 years, whereas mean age for women with high apoE levels (highest quartile) was 57.2 ±
Table 4 – Clinical and laboratory characterization (mean ± SD) and comparisonsa among the HR2 men as a function of CVD events.
3.2. ApoE levels in multivariable risk models in female and male parent study populations
Parameter
To assess potential effects of apoE levels on risk in the parent (BG plus HR1 plus HR2) female and male study groups, apoE quartile analysis using Cox regression was performed. Results demonstrated apoE levels as a significant predictor of risk with increasing quartiles of apoE concentration in both women (Q2/Q1, P = .14; Q3/Q1, P = .032; and Q4/Q1, P = .0002) and men (Q2/Q1, P = .60; Q3/Q1, P = .068; and Q4/Q1, P = .023). However, only in women did apoE dichotomized as highest quartile vs combined lowest 3 quartiles remain predictive of risk in age-adjusted analysis (P = .033).
Clinical Age (y) BMI (kg/m2) Hypertension (%) Metabolic syndrome (%) Alcohol use (%) Smoking status (%) Never Former Current Biomarkers CRP (mg/L) b HDL-C (mmol/L) ApoE (g/L) ApoA1 (g/L) Triglycerides (mmol/L) ApoB (g/L) Cholesterol (mmol/L) LDL-C (mmol/L) Glucose (mmol/L) Creatinine (μmol/L) UAE (mg/24 h) b
3.3. ApoE levels in multivariable risk models within subgroups Preliminary to assessing a potential role for apoE levels in predicting risk within the HR2 subgroups, risk potential of clinical and biomarker parameters was explored by comparison of means for subjects in the HR2 subgroups without and with outcome events in women (Table 3) and in men (Table 4). The HR2 subjects with events in comparison to the subjects without events demonstrated the following: for women—older age; greater BMI; more hypertension; and higher levels of apoE, triglycerides, apoB, total cholesterol,
a
HR2 men without events (n = 347)
HR2 men with events (n = 27)
53.9 ± 12.9 26.5 ± 3.9 65.8 8.0 43.6
63.3 ± 9.3 26.3 ± 4.0 85.2 3.7 55.6
16.1 41.8 42.1
7.4 59.3 33.3
3.58 (2.52-5.15) 1.40 ± 0.24 0.0427 ± 0.0165 1.47 ± 0.23 1.33 ± 0.67 1.11 ± 0.28 5.96 ± 1.14 3.95 ± 1.03 5.05 ± 1.34 92.1 ± 15.3 11.7 (7.4-27.0)
4.15 (2.61-6.89) 1.45 ± 0.28 0.0405 ± 0.0120 1.47 ± 018 1.20 ± 0.65 1.13 ± 0.25 6.10 ± 1.01 4.10 ± 1.01 5.02 ± 0.84 100.4 ± 33.5 20.7 (9.6-65.5)
P
<.001 .65 .039 .42 .23 .18
.13 .48 .60 .80 .23 .62 .53 .55 .81 .087 .035
Mann-Whitney U test. For CRP and UAE, median values and interquartile ranges are given.
b
1000
M ET ABOL I S M CL IN I CA L A N D EX PE R IM EN TA L 6 1 ( 2 0 12 ) 99 6 –1 0 02
Table 5 – Cox regression results for dichotomized apoE levels (highest quartile vs combined 3 lowest quartiles) in the HR1 and HR2 subgroups for women and for men. Subgroup
Women BG (univariate) BG (multivariable) a HR1 (univariate) HR1 (multivariable) b HR2 (univariate) HR2 (multivariable) c Men BG (univariate) BG (multivariable) d HR1 (univariate) HR1 (multivariable) e HR2 (univariate) HR2 (multivariable) f
n
No. of Cox regression results outcome P events Hazard 95% CI ratio
2232 2232 879 879
23 23 31 31
0.45 0.30 2.86 1.43
0.13-1.53 0.09-1.03 1.38-5.94 0.63-3.24
.20 .056 .0046 .39
508 508
15 15
8.3 4.52
2.20-31.28 .0018 1.07-19.12 .040
1813 1813 764 764
52 52 77 77
0.71 0.49 1.02 0.85
0.36-1.42 0.24-0.99 0.60-1.75 0.44-1.63
.34 .047 .93 .62
374 374
27 27
0.57 0.40
0.20-1.66 0.13-1.25
.30 .12
tion (r = 0.47, P < .0001), whereas for HR2 men, results revealed small and nonsignificant correlation (r = 0.04, P = .43). Moreover, to determine whether this relationship had relevance in the context of risk, outcome event maps were generated for women (Fig. 1A) and men (Fig. 1B). The plots demonstrated estimated event rates over the apoE/age domain. For women, the plot shows localization of risk for older subjects at high apoE levels, whereas for men, the plot shows localization of risk for older subjects at low apoE levels. To further explore in women the role of CRP in apoEassociated risk in the setting of high HDL-C, univariate Cox regression was performed for subjects with high HDL-C levels (≥median HDL-C level) and low CRP levels (
a
Adjusted for age, BMI, hypertension, cholesterol, LDL-C, apoB, and triglycerides. b Adjusted for age, BMI, hypertension, cholesterol, HDL-C, triglycerides, and glucose. c Adjusted for age, BMI, hypertension, cholesterol, LDL-C, apoB, triglycerides, and glucose. d Adjusted for age, hypertension, UAE, creatinine, cholesterol, LDL-C, apoB, and triglycerides. e Adjusted for age, hypertension, smoking, cholesterol, HDL-C, LDL-C, apoB, triglycerides, glucose, CRP, and UAE. f Adjusted for age, hypertension, triglycerides, CRP, creatinine, and UAE.
11.4 years. Correlation and regression analyses were used to demonstrate the relationship between apoE levels and age. Results revealed for HR2 women significant positive correla-
4.
The goal of the current study was to investigate a potential role for apoE levels in CVD risk in human subjects likely to display dysfunctional transformation of HDL. Candidate female and male subgroups (HR2) were previously identified based upon defining characteristics of concurrently high levels of HDL-C and CRP [21,22]. Multivariable modeling demonstrated significant association of apoE levels with risk in the female HR2 subgroup but not in the male HR2 subgroup. Further studies on the parent populations also revealed risk with apoE levels that was especially pronounced in women. Our finding of apoE-associated risk is consistent with studies of other workers [33], although it was somewhat unexpected, as apoE levels more generally have been thought to be
400
300
300
ApoE Rank
ApoE Rank
Discussion
200
200
100 100
30
A
40
50
Age (Years)
60
70
30
B
40
50
60
70
Age (Years)
Fig. 1 – Contour outcome event maps in the HR2 subgroup for (A) women and (B) men. The maps give estimated outcome event rates (percentage) as a function of age and rank of apoE level. The color scales for outcome event rates are as follows: for women (A)—dark red, 8%; red, 6%; orange, 5%; yellow, 4%; light green, 3%; dark green, 1%; and for men (B)—dark red, 35%; red, 27%; orange, 22%; yellow, 15%; light green, 10%; dark green, 3%.
M ET ABOL I S M CL IN I CA L A N D EX PE R IM EN TA L 6 1 ( 2 0 1 2) 9 96 –1 0 02
atheroprotective [34]. In addition, our results extended the observation of apoE-associated risk to a younger age group than reported previously [33]. The nonconcordance of apoE-associated risk observed in HR2 women but not in HR2 men may in part be based on the low event rate in HR2 women with low apoE levels as compared with the high event rates of both HR2 women with high apoE levels and HR2 men independent of apoE levels. This notion was supported by correlation and regression results of apoE level with age that demonstrated significant positive correlation in HR2 women but not in HR2 men. Sex differences of apoE-associated risk were further underscored by outcome event mapping results showing that risk in HR2 women was localized to older subjects with higher apoE levels, whereas risk in HR2 men was localized to older subjects with lower apoE levels. Apolipoprotein E is generally believed to be protective against development of atherosclerosis; however, overabundance of apoE has also been associated with increased risk [22,33-36]. This association has been attributed to inhibition of lipoprotein lipolysis and/or activation of very low-density lipoprotein production by apoE, both of which could lead to generation of atherogenic triglyceride-rich lipoprotein remnant particles [25,37]. It may be the case that such processes are exacerbated by the inflammatory setting defining the HR2 subgroup. Indeed, additional results of the current study tend to support the role of inflammatory processes in this regard as demonstrated by the finding of lack of apoEassociated risk in female subjects with high HDL-C levels and low CRP levels. Study limitations included lack of direct characterizations of pathophysiological processes underlying the association of apoE levels with risk, most notably lipolysis of triglyceride-rich lipoprotein and inflammation and oxidative stress. In addition, no direct data were provided relating to the primacy of high HDL-C/high CRP levels in the establishment of risk in HR2. Indeed, characteristics of the subgroup could suggest establishment of risk as related to processes associated with more traditional clinical and biomarker risk variables (age, BMI, cholesterol, LDL-C, apoB, and triglycerides). In the same vein, there was no direct evidence attesting to actual dysfunctional transformation of HDL especially as related to apoE levels. This issue could be approached in future studies by determination of apoE levels in HDL fractions. But more generally, it would be crucial for future studies to provide functional assessments of HDL and physicochemical characterization of HDL to elucidate the nature of potential dysfunctional transformation. Furthermore, additional clinical, biomarker, and polymorphism risk variables should be determined in the study population as well as replication of findings in other populations. In conclusion, the current study investigated the role of apoE levels in incident CVD risk in female and male subjects likely to display dysfunctional transformation of HDL based upon the defining characteristics of the subgroups as having concurrently high levels of HDL-C and CRP. Results revealed risk with apoE levels in the female subgroup but not in the male subgroup. Future studies should be oriented toward elucidation of mechanisms related to roles of apoE in HDL
1001
function and dysfunction especially as related to atherogenesis and CVD risk.
Funding Grant 2001.005 (Netherlands Heart Foundation).
Acknowledgment Contributions of RT Gansevoort, RPF Dullaart, and GJ Navis were made on behalf of the PREVEND study group. We are indebted to all PREVEND collaborators.
Conflict of Interest No potential conflicts of interest for any of the authors.
REFERENCES
[1] Florentin M, Liberopoulos EN, Wierzbicki AS, et al. Multiple actions of high-density lipoprotein. Curr Opin Cardiol 2008;23: 370-8. [2] deGoma EM, deGoma RL, Rader DJ. Beyond high-density lipoprotein cholesterol levels. J Am CollCardiol 2008;51: 2199-211. [3] Sviridov D, Mukhamedova N, Remaley AT, et al. Antiatherogenic functionality of high density lipoprotein: how much versus how good. J Atheroscler Thromb 2008;15:52-62. [4] Ragbir S, Farmer JA. Dysfunctional high-density lipoprotein and atherosclerosis. Curr Atheroscler Rep 2010;12:343-8. [5] Shao B, Oda M, Oram JF, et al. Myeloperoxidase: an oxidative pathway for generating dysfunctional high-density lipoprotein. Chem Res Toxicol 2010;23:447-54. [6] Ansell BJ, Fonarow GC, Fogelman AM. The paradox of dysfunctional high-density lipoprotein. Curr Opin Lipidol 2007;18:427-34. [7] Kontush A, Chapman MJ. Functionally defective high-density lipoprotein: a new therapeutic target at the crossroads of dyslipidemia, inflammation, and atherosclerosis. Pharmacol Rev 2006;58:342-74. [8] Onat A, Hergene G. Low-grade inflammation, and dysfunction of high-density lipoprotein and its apolipoproteins as a major driver of cardiovascular risk. Metab Clin Exp 2011;60: 499-512. [9] De Backer G, de Bacquer D, Kornitzer M. Epidemiological aspects of high density lipoprotein cholesterol. Atherosclerosis 1998;137(Suppl):S1-6. [10] Jeppesen J, Hein HO, Suadicani P, et al. Triglyceride concentration and ischemic heart disease. Circulation 1998;97: 1029-36. [11] von Eckardstein A, Schulte H, Assmann G. Increased risk of myocardial infarction in men with both hypertriglyceridemia and elevated HDL cholesterol. Circulation 1999;99:1925. [12] Bittner V, Simon JA, Fong J, et al. Correlates of high HDL cholesterol among women with coronary heart disease. Am Heart J 2000;139:288-96. [13] Shimoni N, Kaplan M, Keidar S. Cardiovascular diseases in patients with high levels of plasma high density lipoprotein: association with increased plasma oxidative state. Israel Med Association J 2003;5:702-5.
1002
M ET ABOL I S M CL IN I CA L A N D EX PE R IM EN TA L 6 1 ( 2 0 12 ) 99 6 –1 0 02
[14] Agerholm-Larsen B, Nordestgaard BG, Steffensen R, et al. Elevated HDL cholesterol is a risk factor for ischemic heart disease in white women when caused by a common mutation in the cholesteryl ester transfer protein gene. Circulation 2000;101:1907-12. [15] Gnasso A, Motti C, Irace C, et al. The ARG allele in position 192 of PON1 is associated with carotid atherosclerosis in subjects with elevated HDLs. Atherosclerosis 2002;164:289-95. [16] deLemos AS, Wolfe ML, Long CJ, et al. Identification of genetic variants in endothelial lipase in persons with elevated highdensity lipoprotein cholesterol. Circulation 2002;106:1321-6. [17] Ma K, Cilingiroglu M, Otvos JD, et al. Endothelial lipase is a major genetic determinant for high-density lipoprotein concentration, structure, and metabolism. PNAS 2003;100: 2748-53. [18] Broedl UC, Jin W, Rader DJ. Endothelial lipase: a modulator of lipoprotein metabolism upregulated by inflammation. Trends Cardiovasc Med 2004;14:202-6. [19] Grundy SM, Vega GL, Otvos JD, et al. Hepatic lipase activity influences high density lipoprotein subclass distribution in normotriglyceridemic men: genetic and pharmacological evidence. J Lipid Res 1999;40:229-34. [20] Ruel IL, Couture P, Cohn JS, et al. Evidence that hepatic lipase deficiency in humans is not associated with proatherogenic changes in HDL composition and metabolism. J Lipid Res 2004;45:1528-37. [21] Corsetti JP, Gansevoort RT, Sparks CE, et al. HDL protection against primary cardiac risk is lost with inflammation. Eur J Clin Invest 2010;40:483-9. [22] Corsetti JP, Gansevoort RT, Navis GJ, et al. LPL polymorphism (D9N) predicts cardiovascular disease risk directly and through interaction with CETP polymorphism (TaqIB) in women with high HDL cholesterol and CRP. Atherosclerosis 2011;214:373-6. [23] Corsetti JP, Ryan D, Rainwater DL, et al. CETP polymorphism (TaqIB) associates with risk in postinfarction patients with high HDL cholesterol and high CRP levels. Arterioscler Thromb Vasc Biol 2010;30:1657-64. [24] Corsetti JP, Ryan D, Moss AJ, et al. NAD(P)H oxidase polymorphism (C242T) and high HDL-C associate with recurrent coronary events in post infarction patients. Atherosclerosis 2007;196:461-8. [25] Huang Y. Mechanisms linking apolipoprotein E isoforms with cardiovascular and neurological diseases. Curr Opin Lipidol 2010;21:337-45.
[26] Greenow K, Pearce NJ, Ramji DP. The key role of apolipoprotein E in atherosclerosis. J Mol Med 2005;83:329-42. [27] Jofre-Monseny L, Minihane AM, Rimbach G. Impact of apoE genotype on oxidative stress, inflammation and disease risk. Mol Nutr Food Res 2008;52:131-45. [28] Pinto-Sietsma SJ, Janssen WMT, Hillege HL, et al. Urinary albumin excretion is associated with renal functional abnormalities in a non-diabetic population. J Am Soc Nephrol 2000;11:1882-8. [29] Lambers Heerspink HJ, Brantsma AH, de Zeeuw D, et al. Albuminuria assessed from first-morning-void urine samples versus 24 hour urine collections as a predictor of cardiovascular morbidity and mortality. Am J Epidemiol 2008;168: 897-905. [30] Verhave JD, Gansevoort RT, Hillege HL, et al. An elevated urinary albumin excretion predicts de novo development of renal function impairment in the general population. Kidney Int 2004;66:S18-21. [31] Borggreve SE, Hillege HL, Wolffenbuttel BHR, et al. An increased coronary risk is paradoxically associated with common cholesteryl ester transfer protein gene variations that relate to higher high-density lipoprotein cholesterol: a population-based study. J Clin Endocrinol Metab 2006;91: 3382-8. [32] Corsetti JP, Zareba W, Moss AJ, et al. Serum glucose and triglyceride determine high-risk subgroups in nondiabetic postinfarction patients. Atherosclerosis 2005;183: 293-300. [33] Mooijaart SP, Berbee JFP, van Heemst D, et al. ApoE plasma levels and risk of cardiovascular mortality in old age. PLoS Med 2006;3:874-83. [34] Reilly M, Rader DJ. Apolipoprotein E and coronary disease: a puzzling paradox. PLoS Med 2006;3:736-8. [35] Soderlund S, Watanabe H, Ehnholm C, Jauhiainen M, et al. Increased apolipoprotein E level and reduced high-density lipoprotein mean particle size associate with low highdensity lipoprotein cholesterol and features of metabolic syndrome. Metab Clin Exp 2010;60:1502-9. [36] Mahley RW, Weisgraber KH, Huang Y. Apolipoprotein E: structure determines function, from atherosclerosis to Alzheimer’s disease to AIDS. J Lipid Res 2009;50: S183-8. [37] Huang Y, Liu XQ, Rall SC, et al. Overexpression and accumulation of apolipoprotein E as a cause of hypertriglyceridemia. J Biol Chem 1998;273:26388-93.
Available online at www.sciencedirect.com
Metabolism www.metabolismjournal.com
Apolipoprotein E predicts incident cardiovascular disease risk in women but not in men with concurrently high levels of high-density lipoprotein cholesterol and C-reactive protein James P. Corsetti a,⁎, Ron T. Gansevoort b , Stephan J.L. Bakker b , GerJan Navis b , Charles E. Sparks a , Robin P.F. Dullaart c a b c
Department of Pathology and Laboratory Medicine, University of Rochester School of Medicine and Dentistry, Rochester, NY 14642, USA Department of Nephrology, University of Groningen and University Medical Center Groningen, 9700 RB Groningen, the Netherlands Department of Endocrinology, University of Groningen and University Medical Center Groningen, 9700 RB Groningen, the Netherlands
A R T I C LE I N FO
AB S T R A C T
Article history:
Although there is great interest in the notion that dysfunctional transformation of high-
Received 13 October 2011
density lipoprotein (HDL) facilitates development of atherosclerosis and cardiovascular
Accepted 19 November 2011
disease (CVD), few studies in human populations directly address this issue. As apolipoprotein E (apoE) is a constituent of HDL thought to be important for HDL antiatherogenic function, we sought to assess the role of apoE in CVD risk in subjects likely to display dysfunctional transformation of HDL. Association of apoE levels with incident CVD risk was investigated using Cox multivariable proportional hazards modeling. Analyses were performed in subgroups of women and men likely to display dysfunctional transformation of HDL deriving from previous subgroup identification based upon defining characteristics of concurrently high levels of HDL cholesterol and systemic inflammation as reflected by high C-reactive protein levels. Results revealed apoE levels (dichotomized as highest quartile vs combined 3 lowest quartiles) as predicting subgroup risk in women (hazard ratio, 4.52; 95% confidence interval, 1.07-19.12; P = .040) but not in men. Further sex differences were manifested in terms of the relationship of apoE levels with age. Analysis revealed positive correlation of apoE levels with age in women (r = 0.47, P < .0001) but not in men (r = 0.04, P = .43). Apolipoprotein E levels predict incident CVD risk in women with high levels of HDL cholesterol and C-reactive protein but not in men. Future studies should be oriented toward investigations of apoE as related to multiplicity of HDL functionality and toward assessment of potential roles for apoE in dysfunctional transformation of HDL. © 2012 Elsevier Inc. All rights reserved.
1.
Introduction
In addition to the traditional focus on high-density lipoprotein (HDL) levels, there is growing interest in the role of HDL
functionality in relation to atherogenesis and cardiovascular disease (CVD). In this context, HDL antiatherogenic functionality is thought to be based upon multiple protective roles including cholesterol clearance by reverse cholesterol transport (RCT),
Author contributions: JPC: study conception and design, data analysis and interpretation, manuscript drafting; RTG: data acquisition, manuscript editing; SJLB: data acquisition, manuscript editing; GJN: data acquisition, manuscript editing; CES: study conception and design, data analysis and interpretation, manuscript editing; RPFD: study conception and design, data analysis and interpretation, manuscript editing. ⁎ Corresponding author. Department of Pathology and Laboratory Medicine, University of Rochester Medical Center, Box 626, Rochester, NY 14642, USA. Tel.: +1 585 275 4907; fax: +1 585 273 3003. E-mail address: [email protected] (J.P. Corsetti). 0026-0495/$ – see front matter © 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.metabol.2011.11.010
M ET ABOL I S M CL IN I CA L A N D EX PE R IM EN TA L 6 1 ( 2 0 1 2) 9 96 –1 0 02
resistance to vascular inflammatory and oxidative stress processes, preservation of endothelial function, and antithrombotic effects [1-3]. It is now widely recognized that HDL particles can become dysfunctional in the setting of inflammation and oxidative stress [4-8]. Such considerations are especially important in view of intense efforts currently under way to reduce residual CVD risk by raising HDL cholesterol (HDL-C) levels. However, despite potential implications of dysfunctional transformation of HDL in regard to CVD risk, little work is available addressing this issue in human populations. To this end, we have studied associations of CVD risk in humans with high HDL-C levels in the setting of inflammation as reflected by high C-reactive protein (CRP) levels. Subjects with high HDL-C levels were chosen for the study, as risk associated with dysfunctional HDL transformation would be difficult to discern in subjects with low HDL-C levels given the wellknown inverse correlation with risk. In addition, it should be noted that high levels of HDL-C have been associated with risk under various circumstances as reported in earlier studies [920]. Thus, for subjects with relatively high levels of both HDL-C and CRP, we have identified high risk in men and women without previous history of CVD in the Prevention of Renal and Vascular End-Stage Disease (PREVEND) study group [21,22] as well as in postinfarction patients in the Thrombogenic Factors and Recurrent Coronary Events Postinfarction study group [23]. Moreover, these studies used functional genetic polymorphisms of relevant genes to provide evidence in support of involvement in risk of oxidative stress (CYBA) [24] and early phases of RCT (CETP, LPL) [22,23]. Antiatherogenic properties of HDL are thought to derive from multiple HDL particle constituents including apolipoproteins (apoA1, apoA2, and apoE) and enzymes (paraoxonase 1, lecithin-cholesterol acyltransferase, and glutathione peroxidase) [7]. With particular regard to apoE, antiatherogenic properties are usually ascribed to effects on lipoprotein particle clearance in conjunction with RCT [25,26]. However, in addition to its role in RCT, evidence is accumulating that apoE has significant immunoregulatory properties that could also play a role in apoE-associated antiatherogenic function [27]. In view of the multiple potential protective roles for apoE in terms of atherogenesis and CVD, we assessed effects on incident cardiovascular disease risk of plasma apoE levels in previously identified high-risk subgroups of men and women [21,22] defined by concurrently high levels of HDL-C and CRP from the PREVEND population study [28].
2.
Subjects and methods
2.1.
Study populations
The current work was based on PREVEND [28], a prospective longitudinal study of albuminuria in predicting cardiovascular [29] and renal disease [30]. PREVEND was approved by the medical ethics committee of the University of Groningen, the Netherlands; informed consent was obtained from all subjects. Preliminary study exclusions were for insulin-using diabetic patients and pregnancy. Further exclusions for the current study included history of diabetes mellitus, renal disease, previous CVD, incomplete laboratory results, and CRP
997
levels of at least 10 mg/L (to avoid confounding by intercurrent illness). Outcome events included CVD mortality and CVDrelated hospitalization (acute myocardial infarction, acute and subacute ischemic heart disease, coronary artery bypass grafting, and percutaneous transluminal coronary angioplasty). Median follow-up time was 7.6 years. Detailed descriptions of definitions and data acquisition techniques were reported previously [21,28-31].
2.2.
Blood biomarkers
Serum and plasma biomarkers were analyzed on samples stored at −20°C prepared from venous blood after overnight fast with collection after 15 minutes of rest. Total cholesterol, HDL-C, triglycerides, apoA1, apoB, glucose, high-sensitivity CRP, creatinine, and urine albumin were determined as described previously [21,31]. Urinary albumin excretion rate (UAE) is the mean of two 24-hour collections. Apolipoprotein E levels were determined by immunonephelometry (BNII; Dade Behring, Marburg, Germany). Low-density lipoprotein cholesterol (LDL-C) levels were estimated using the Friedewald equation.
2.3.
Statistical analyses
Analyses were performed using Statistica 10.0 (StatSoft, Tulsa, OK) including Mann-Whitney U, Kruskal-Wallis with Bonferroni correction, and Cox proportional hazards multivariable regression with adjustment for clinical covariates (age, hypertension, body mass index [BMI], ethanol use, and smoking) and urine and blood biomarkers (UAE, creatinine, apoA1, apoB, cholesterol, CRP, HDL-C, triglycerides, LDL-C, and glucose) based on significance (P < .10) in univariate analysis. This approach was taken to minimize inclusion of potentially extraneous predictor variables in multivariable models. Apolipoprotein E was treated as a dichotomized variable (highest quartile vs combined lowest 3 quartiles). Statistical significance was at the P < .05 level. Outcome event mapping [22,24,32], a graphical exploratory data analysis tool, was used to identify high-risk subgroups as a function of 2 risk parameters. In this case, risk parameters were HDL-C and CRP levels. A 3-dimensional scatter plot is generated with CVD outcome on the zaxis (coded 0 for no outcome and +1 for outcome) vs the 2 rank-transformed (to more evenly distribute patients over the bivariate risk domain) risk parameters (x-axis, HDL-C levels; y-axis, CRP levels). Application of a smoothing algorithm results in a surface (outcome event map) generally demonstrating peaks and valleys and where height above the bivariate x-y plane approximates the outcome rate. Subjects contained within peaks in the mappings comprise high-risk subgroups. Peaks are defined as regions with outcome rates greater than the mean outcome rate of the total population. In practice, a contour line representing the constant value of the mean outcome rate is generated on the mapping. In the present case, the mean outcome rate was 1.9% for women and 5.5% for men. High-risk subjects are then identified as those subjects within the footprint (projection) cast on the x-y plane by the contour line corresponding to the value of the mean outcome rate.
998
M ET ABOL I S M CL IN I CA L A N D EX PE R IM EN TA L 6 1 ( 2 0 12 ) 99 6 –1 0 02
Table 1 – Clinical and laboratory characterization (mean ± SD) and comparisonsa among BG, HR1, and HR2 subjects in female participants. BG (n = 2232)
CVD outcomes % (n) Clinical Age (y) BMI (kg/m2) Hypertension (%) Metabolic syndrome (%) Alcohol use (%) Smoking status (%) Never Former Current Biomarkers CRP (mg/L) c
1.03 (23)
3.53 (31)
2.95 (15)
46.2 ± 11.6 24.2 ± 3.5 27.2
50.1 ± 12.3 28.2 ± 5.0 49.0
48.5 ± 12.9 27.0 ± 4.8 45.3
.02 <.001 .18
3.53
46.29
12.35
<.001
20.2
12.5
17.0
.02
HDL-C (mmol/L) ApoE (g/L) ApoA1 (g/L) Triglycerides (mmol/L) ApoB (g/L) Cholesterol (mmol/L) LDL-C (mmol/L) Glucose (mmol/L) Creatinine (μmol/L) UAE (mg/24 h) c
HR1 (n = 879)
HR2 (n = 508)
Pb
Parameter
.57
.047 35.8 34.3 30.0
31.9 28.4 39.7
31.9 34.1 34.1
0.64 (0.34-1.25) 1.67 ± 0.38
2.43 (1.47-4.31) 1.07 ± 0.16
4.66 (3.35-6.42) 1.55 ± 0.21
<.001
0.0365 ± 0.0128 1.54 ± 0.30 0.99 ± 0.46
0.0422 ± 0.0180 1.28 ± 0.20 1.74 ± 1.03
0.0377 ± 0.0133 1.59 ± 0.25 1.27 ± 0.55
<.001
0.91 ± 0.26 5.47 ± 1.11
1.15 ± 0.33 5.87 ± 1.19
1.01 ± 0.27 5.67 ± 1.17
<.001 .0015
3.35 ± 1.04
4.00 ± 1.04
3.53 ± 1.09
<.001
4.50 ± 0.61
4.98 ± 1.32
4.71 ± 0.85
<.001
75.8 ± 11.8
76.2 ± 12.8
76.3 ± 10.1
.44
7.8 (5.6-12.2)
9.4 (6.2-18.4)
8.9 (5.8-15.1)
.036
<.001
<.001 <.001
ed as HR1); and the other, by high levels of HDL-C and high levels of CRP (henceforth designated as HR2). Lower-risk female subjects and lower-risk male subjects constituted background subgroups (henceforth designated BG). Clinical and laboratory characterization for the subgroups is given in Table 1 for women and in Table 2 for men. Briefly, women in HR1 vs HR2 were slightly older; exhibited more features of metabolic syndrome; used alcohol less; and had higher levels of apoE, cholesterol, LDL-C, triglycerides, glucose, and apoB, and lower levels of HDL-C, CRP, and apoA1. Men in HR1 vs HR2
Table 2 – Clinical and laboratory characterization (mean ± SD) and comparisonsa among BG, HR1, and HR2 subjects in male participants. BG (n = 1813)
CVD outcomes % (n) Clinical Age (y) BMI (kg/m2) Hypertension (%) Metabolic syndrome (%) Alcohol use (%) Smoking status (%) Never Former Current Biomarkers CRP (mg/L) c
2.87 (52)
10.08 (77)
7.22 (27)
46.1 ± 12.1 25.0 ± 3.2 47.9
51.6 ± 12.0 28.0 ± 3.8 64.2
54.6 ± 12.9 26.5 ± 3.9 67.2
<.001 <.001 .31
9.65
53.99
7.67
<.001
34.3
29.4
44.5
<.001
HDL-C (mmol/L) ApoE (g/L)
a
Kruskal-Wallis with Bonferroni correction with results as follows: (1) BG vs HR1—all parameters significantly different except for creatinine; and (2) BG vs HR2—all parameters significantly different except for alcohol use, distribution of smokers, apoE, and creatinine. b Comparison of HR1 vs HR2. c For CRP and UAE, median values and interquartile ranges are given.
ApoA1 (g/L) Triglycerides (mmol/L) ApoB (g/L) Cholesterol (mmol/L) LDL-C (mmol/L) Glucose (mmol/L) Creatinine (μmol/L) UAE (mg/24 h) c
The resultant outcome event maps for the current study have been reported previously [21,22].
3.
Results
3.1.
Study populations
a
Previous results using outcome event mapping, a graphical exploratory data analysis tool [23,31], revealed 2 similar highrisk subgroups in the female [22] and male [21] subcohorts. One subgroup was characterized by relatively low levels of HDL-C and relatively high levels of CRP (henceforth designat-
HR1 (n = 764)
HR2 (n = 374)
Pb
Parameter
.12
.004 34.0 40.4 25.6
18.4 32.9 48.7
15.5 43.1 41.4
0.53 (0.30-0.81) 1.28 ± 0.32
2.79 (1.78-4.73) 0.86 ± 0.13
3.60 (2.53-5.22) 1.40 ± 0.25
<.001
0.0369 ± 0.0151 1.34 ± 0.26 1.34 ± 0.93
0.0461 ± 0.0239 1.15 ± 0.17 2.25 ± 1.78
0.0426 ± 0.0162 1.47 ± 0.23 1.32 ± 0.67
.26 <.001 <.001
1.00 ± 0.27 5.50 ± 1.03
1.21 ± 0.31 5.92 ± 1.26
1.11 ± 0.28 5.97 ± 1.13
<.001 .51
3.60 ± 0.93
4.03 ± 1.18
3.96 ± 1.03
.19
4.75 ± 0.77
5.26 ± 1.47
5.05 ± 1.31
.001
90.2 ± 12.3
93.9 ± 37.2
92.7 ± 17.3
.32
9.0 (6.4-15.5) 12.7 (7.5-30.3) 11.9 (7.4-29.2)
<.001
.44
Kruskal-Wallis with Bonferroni correction with results as follows: (1) BG vs HR1—all parameters significantly different; and (2) BG vs HR2—all parameters significantly different except for metabolic syndrome, HDL/apoA1, triglycerides, apoB/apoA1, and creatinine. b Comparison of HR1 vs HR2. c For CRP and UAE, median values and interquartile ranges are given.
999
M ET ABOL I S M CL IN I CA L A N D EX PE R IM EN TA L 6 1 ( 2 0 1 2) 9 96 –1 0 02
Table 3 – Clinical and laboratory characterization (mean ± SD) and comparisonsa among the HR2 women as a function of CVD events. Parameter
Clinical Age (y) BMI (kg/m2) Hypertension (%) Metabolic syndrome (%) Alcohol use (%) Smoking status (%) Never Former Current Biomarkers CRP (mg/L) b HDL-C (mmol/L) ApoE (g/L) ApoA1 (g/L) Triglycerides (mmol/L) ApoB (g/L) Cholesterol (mmol/L) LDL-C (mmol/L) Glucose (mmol/L) Creatinine (μmol/L) UAE (mg/24 h) b a b
HR2 women HR2 women without events with events (n = 493) (n = 15)
P
48.2 ± 12.8 26.9 ± 4.8 44.2 12.3
60.3 ± 9.6 29.8 ± 5.3 80.0 13.3
<.001 .014 .006 .91
16.9
20.0
31.9 34.7 33.5
33.3 13.3 53.3
4.60 (3.34-6.39) 1.55 ± 0.21 0.0374 ± 0.0131 1.59 ± 0.25 1.26 ± 0.55 1.00 ± 0.27 5.63 ± 1.16 3.50 ± 1.08 4.68 ± 0.79 76.2 ± 10.0 8.8 (5.8-14.9)
5.64 (4.46-7.23) .082 1.55 ± 0.16 .69 0.0505 ± 0.0103 <.001 1.58 ± 0.22 .99 1.63 ± 0.52 .003 1.23 ± 0.27 .002 6.81 ± 1.04 <.001 4.51 ± 0.99 <.001 5.37 ± 2.00 .15 77.7 ± 13.3 .88 13.4 (6.9-25.2) .13
.75 .16
Mann-Whitney U test. For CRP and UAE, median values and interquartile ranges are given.
demonstrated similar differences except for younger age and no differences in apoE, cholesterol, and LDL-C.
and LDL-C; for men—older age, more hypertension, and higher urinary albumin excretion. Results of multivariable modeling of potential associations of apoE levels with risk within the BG, HR1, and HR2 subgroups in women and men are given in Table 5. For HR1 in women, there was a loss of significance of apoE levels in the multivariable model relative to the univariate result, whereas for HR2 in women, apoE levels maintained significance as a predictor of risk in the multivariable model. For men, apoE levels did not significantly predict risk in the HR1 and HR2 subgroups. Analyses were carried out using Cox proportional hazards modeling using dichotomized apoE levels first in univariate Cox analysis followed by multivariable modeling adjusted for significant clinical covariates and blood marker levels as described in “Subjects and Methods.” To investigate the lack of apoE-associated risk in HR2 men in contrast to HR2 women, event rates for the 2 subgroups were compared. The event rate for HR2 men was 7.22%; and as noted, it was not a function of apoE levels. The event rates for HR2 women with low vs high apoE levels were 0.88% and 7.02%, respectively. Thus, the event rate in HR2 women with high apoE levels was almost exactly the same as the event rate in HR2 men; however, the event rate in HR2 women with low apoE levels was close to an order of magnitude smaller. To explore whether age might contribute to the sex difference, apoE levels as a function of age in HR2 women were investigated. Mean age for women with low apoE levels (lowest 3 quartiles) was 45.5 ± 12.1 years, whereas mean age for women with high apoE levels (highest quartile) was 57.2 ±
Table 4 – Clinical and laboratory characterization (mean ± SD) and comparisonsa among the HR2 men as a function of CVD events.
3.2. ApoE levels in multivariable risk models in female and male parent study populations
Parameter
To assess potential effects of apoE levels on risk in the parent (BG plus HR1 plus HR2) female and male study groups, apoE quartile analysis using Cox regression was performed. Results demonstrated apoE levels as a significant predictor of risk with increasing quartiles of apoE concentration in both women (Q2/Q1, P = .14; Q3/Q1, P = .032; and Q4/Q1, P = .0002) and men (Q2/Q1, P = .60; Q3/Q1, P = .068; and Q4/Q1, P = .023). However, only in women did apoE dichotomized as highest quartile vs combined lowest 3 quartiles remain predictive of risk in age-adjusted analysis (P = .033).
Clinical Age (y) BMI (kg/m2) Hypertension (%) Metabolic syndrome (%) Alcohol use (%) Smoking status (%) Never Former Current Biomarkers CRP (mg/L) b HDL-C (mmol/L) ApoE (g/L) ApoA1 (g/L) Triglycerides (mmol/L) ApoB (g/L) Cholesterol (mmol/L) LDL-C (mmol/L) Glucose (mmol/L) Creatinine (μmol/L) UAE (mg/24 h) b
3.3. ApoE levels in multivariable risk models within subgroups Preliminary to assessing a potential role for apoE levels in predicting risk within the HR2 subgroups, risk potential of clinical and biomarker parameters was explored by comparison of means for subjects in the HR2 subgroups without and with outcome events in women (Table 3) and in men (Table 4). The HR2 subjects with events in comparison to the subjects without events demonstrated the following: for women—older age; greater BMI; more hypertension; and higher levels of apoE, triglycerides, apoB, total cholesterol,
a
HR2 men without events (n = 347)
HR2 men with events (n = 27)
53.9 ± 12.9 26.5 ± 3.9 65.8 8.0 43.6
63.3 ± 9.3 26.3 ± 4.0 85.2 3.7 55.6
16.1 41.8 42.1
7.4 59.3 33.3
3.58 (2.52-5.15) 1.40 ± 0.24 0.0427 ± 0.0165 1.47 ± 0.23 1.33 ± 0.67 1.11 ± 0.28 5.96 ± 1.14 3.95 ± 1.03 5.05 ± 1.34 92.1 ± 15.3 11.7 (7.4-27.0)
4.15 (2.61-6.89) 1.45 ± 0.28 0.0405 ± 0.0120 1.47 ± 018 1.20 ± 0.65 1.13 ± 0.25 6.10 ± 1.01 4.10 ± 1.01 5.02 ± 0.84 100.4 ± 33.5 20.7 (9.6-65.5)
P
<.001 .65 .039 .42 .23 .18
.13 .48 .60 .80 .23 .62 .53 .55 .81 .087 .035
Mann-Whitney U test. For CRP and UAE, median values and interquartile ranges are given.
b
1000
M ET ABOL I S M CL IN I CA L A N D EX PE R IM EN TA L 6 1 ( 2 0 12 ) 99 6 –1 0 02
Table 5 – Cox regression results for dichotomized apoE levels (highest quartile vs combined 3 lowest quartiles) in the HR1 and HR2 subgroups for women and for men. Subgroup
Women BG (univariate) BG (multivariable) a HR1 (univariate) HR1 (multivariable) b HR2 (univariate) HR2 (multivariable) c Men BG (univariate) BG (multivariable) d HR1 (univariate) HR1 (multivariable) e HR2 (univariate) HR2 (multivariable) f
n
No. of Cox regression results outcome P events Hazard 95% CI ratio
2232 2232 879 879
23 23 31 31
0.45 0.30 2.86 1.43
0.13-1.53 0.09-1.03 1.38-5.94 0.63-3.24
.20 .056 .0046 .39
508 508
15 15
8.3 4.52
2.20-31.28 .0018 1.07-19.12 .040
1813 1813 764 764
52 52 77 77
0.71 0.49 1.02 0.85
0.36-1.42 0.24-0.99 0.60-1.75 0.44-1.63
.34 .047 .93 .62
374 374
27 27
0.57 0.40
0.20-1.66 0.13-1.25
.30 .12
tion (r = 0.47, P < .0001), whereas for HR2 men, results revealed small and nonsignificant correlation (r = 0.04, P = .43). Moreover, to determine whether this relationship had relevance in the context of risk, outcome event maps were generated for women (Fig. 1A) and men (Fig. 1B). The plots demonstrated estimated event rates over the apoE/age domain. For women, the plot shows localization of risk for older subjects at high apoE levels, whereas for men, the plot shows localization of risk for older subjects at low apoE levels. To further explore in women the role of CRP in apoEassociated risk in the setting of high HDL-C, univariate Cox regression was performed for subjects with high HDL-C levels (≥median HDL-C level) and low CRP levels (
a
Adjusted for age, BMI, hypertension, cholesterol, LDL-C, apoB, and triglycerides. b Adjusted for age, BMI, hypertension, cholesterol, HDL-C, triglycerides, and glucose. c Adjusted for age, BMI, hypertension, cholesterol, LDL-C, apoB, triglycerides, and glucose. d Adjusted for age, hypertension, UAE, creatinine, cholesterol, LDL-C, apoB, and triglycerides. e Adjusted for age, hypertension, smoking, cholesterol, HDL-C, LDL-C, apoB, triglycerides, glucose, CRP, and UAE. f Adjusted for age, hypertension, triglycerides, CRP, creatinine, and UAE.
11.4 years. Correlation and regression analyses were used to demonstrate the relationship between apoE levels and age. Results revealed for HR2 women significant positive correla-
4.
The goal of the current study was to investigate a potential role for apoE levels in CVD risk in human subjects likely to display dysfunctional transformation of HDL. Candidate female and male subgroups (HR2) were previously identified based upon defining characteristics of concurrently high levels of HDL-C and CRP [21,22]. Multivariable modeling demonstrated significant association of apoE levels with risk in the female HR2 subgroup but not in the male HR2 subgroup. Further studies on the parent populations also revealed risk with apoE levels that was especially pronounced in women. Our finding of apoE-associated risk is consistent with studies of other workers [33], although it was somewhat unexpected, as apoE levels more generally have been thought to be
400
300
300
ApoE Rank
ApoE Rank
Discussion
200
200
100 100
30
A
40
50
Age (Years)
60
70
30
B
40
50
60
70
Age (Years)
Fig. 1 – Contour outcome event maps in the HR2 subgroup for (A) women and (B) men. The maps give estimated outcome event rates (percentage) as a function of age and rank of apoE level. The color scales for outcome event rates are as follows: for women (A)—dark red, 8%; red, 6%; orange, 5%; yellow, 4%; light green, 3%; dark green, 1%; and for men (B)—dark red, 35%; red, 27%; orange, 22%; yellow, 15%; light green, 10%; dark green, 3%.
M ET ABOL I S M CL IN I CA L A N D EX PE R IM EN TA L 6 1 ( 2 0 1 2) 9 96 –1 0 02
atheroprotective [34]. In addition, our results extended the observation of apoE-associated risk to a younger age group than reported previously [33]. The nonconcordance of apoE-associated risk observed in HR2 women but not in HR2 men may in part be based on the low event rate in HR2 women with low apoE levels as compared with the high event rates of both HR2 women with high apoE levels and HR2 men independent of apoE levels. This notion was supported by correlation and regression results of apoE level with age that demonstrated significant positive correlation in HR2 women but not in HR2 men. Sex differences of apoE-associated risk were further underscored by outcome event mapping results showing that risk in HR2 women was localized to older subjects with higher apoE levels, whereas risk in HR2 men was localized to older subjects with lower apoE levels. Apolipoprotein E is generally believed to be protective against development of atherosclerosis; however, overabundance of apoE has also been associated with increased risk [22,33-36]. This association has been attributed to inhibition of lipoprotein lipolysis and/or activation of very low-density lipoprotein production by apoE, both of which could lead to generation of atherogenic triglyceride-rich lipoprotein remnant particles [25,37]. It may be the case that such processes are exacerbated by the inflammatory setting defining the HR2 subgroup. Indeed, additional results of the current study tend to support the role of inflammatory processes in this regard as demonstrated by the finding of lack of apoEassociated risk in female subjects with high HDL-C levels and low CRP levels. Study limitations included lack of direct characterizations of pathophysiological processes underlying the association of apoE levels with risk, most notably lipolysis of triglyceride-rich lipoprotein and inflammation and oxidative stress. In addition, no direct data were provided relating to the primacy of high HDL-C/high CRP levels in the establishment of risk in HR2. Indeed, characteristics of the subgroup could suggest establishment of risk as related to processes associated with more traditional clinical and biomarker risk variables (age, BMI, cholesterol, LDL-C, apoB, and triglycerides). In the same vein, there was no direct evidence attesting to actual dysfunctional transformation of HDL especially as related to apoE levels. This issue could be approached in future studies by determination of apoE levels in HDL fractions. But more generally, it would be crucial for future studies to provide functional assessments of HDL and physicochemical characterization of HDL to elucidate the nature of potential dysfunctional transformation. Furthermore, additional clinical, biomarker, and polymorphism risk variables should be determined in the study population as well as replication of findings in other populations. In conclusion, the current study investigated the role of apoE levels in incident CVD risk in female and male subjects likely to display dysfunctional transformation of HDL based upon the defining characteristics of the subgroups as having concurrently high levels of HDL-C and CRP. Results revealed risk with apoE levels in the female subgroup but not in the male subgroup. Future studies should be oriented toward elucidation of mechanisms related to roles of apoE in HDL
1001
function and dysfunction especially as related to atherogenesis and CVD risk.
Funding Grant 2001.005 (Netherlands Heart Foundation).
Acknowledgment Contributions of RT Gansevoort, RPF Dullaart, and GJ Navis were made on behalf of the PREVEND study group. We are indebted to all PREVEND collaborators.
Conflict of Interest No potential conflicts of interest for any of the authors.
REFERENCES
[1] Florentin M, Liberopoulos EN, Wierzbicki AS, et al. Multiple actions of high-density lipoprotein. Curr Opin Cardiol 2008;23: 370-8. [2] deGoma EM, deGoma RL, Rader DJ. Beyond high-density lipoprotein cholesterol levels. J Am CollCardiol 2008;51: 2199-211. [3] Sviridov D, Mukhamedova N, Remaley AT, et al. Antiatherogenic functionality of high density lipoprotein: how much versus how good. J Atheroscler Thromb 2008;15:52-62. [4] Ragbir S, Farmer JA. Dysfunctional high-density lipoprotein and atherosclerosis. Curr Atheroscler Rep 2010;12:343-8. [5] Shao B, Oda M, Oram JF, et al. Myeloperoxidase: an oxidative pathway for generating dysfunctional high-density lipoprotein. Chem Res Toxicol 2010;23:447-54. [6] Ansell BJ, Fonarow GC, Fogelman AM. The paradox of dysfunctional high-density lipoprotein. Curr Opin Lipidol 2007;18:427-34. [7] Kontush A, Chapman MJ. Functionally defective high-density lipoprotein: a new therapeutic target at the crossroads of dyslipidemia, inflammation, and atherosclerosis. Pharmacol Rev 2006;58:342-74. [8] Onat A, Hergene G. Low-grade inflammation, and dysfunction of high-density lipoprotein and its apolipoproteins as a major driver of cardiovascular risk. Metab Clin Exp 2011;60: 499-512. [9] De Backer G, de Bacquer D, Kornitzer M. Epidemiological aspects of high density lipoprotein cholesterol. Atherosclerosis 1998;137(Suppl):S1-6. [10] Jeppesen J, Hein HO, Suadicani P, et al. Triglyceride concentration and ischemic heart disease. Circulation 1998;97: 1029-36. [11] von Eckardstein A, Schulte H, Assmann G. Increased risk of myocardial infarction in men with both hypertriglyceridemia and elevated HDL cholesterol. Circulation 1999;99:1925. [12] Bittner V, Simon JA, Fong J, et al. Correlates of high HDL cholesterol among women with coronary heart disease. Am Heart J 2000;139:288-96. [13] Shimoni N, Kaplan M, Keidar S. Cardiovascular diseases in patients with high levels of plasma high density lipoprotein: association with increased plasma oxidative state. Israel Med Association J 2003;5:702-5.
1002
M ET ABOL I S M CL IN I CA L A N D EX PE R IM EN TA L 6 1 ( 2 0 12 ) 99 6 –1 0 02
[14] Agerholm-Larsen B, Nordestgaard BG, Steffensen R, et al. Elevated HDL cholesterol is a risk factor for ischemic heart disease in white women when caused by a common mutation in the cholesteryl ester transfer protein gene. Circulation 2000;101:1907-12. [15] Gnasso A, Motti C, Irace C, et al. The ARG allele in position 192 of PON1 is associated with carotid atherosclerosis in subjects with elevated HDLs. Atherosclerosis 2002;164:289-95. [16] deLemos AS, Wolfe ML, Long CJ, et al. Identification of genetic variants in endothelial lipase in persons with elevated highdensity lipoprotein cholesterol. Circulation 2002;106:1321-6. [17] Ma K, Cilingiroglu M, Otvos JD, et al. Endothelial lipase is a major genetic determinant for high-density lipoprotein concentration, structure, and metabolism. PNAS 2003;100: 2748-53. [18] Broedl UC, Jin W, Rader DJ. Endothelial lipase: a modulator of lipoprotein metabolism upregulated by inflammation. Trends Cardiovasc Med 2004;14:202-6. [19] Grundy SM, Vega GL, Otvos JD, et al. Hepatic lipase activity influences high density lipoprotein subclass distribution in normotriglyceridemic men: genetic and pharmacological evidence. J Lipid Res 1999;40:229-34. [20] Ruel IL, Couture P, Cohn JS, et al. Evidence that hepatic lipase deficiency in humans is not associated with proatherogenic changes in HDL composition and metabolism. J Lipid Res 2004;45:1528-37. [21] Corsetti JP, Gansevoort RT, Sparks CE, et al. HDL protection against primary cardiac risk is lost with inflammation. Eur J Clin Invest 2010;40:483-9. [22] Corsetti JP, Gansevoort RT, Navis GJ, et al. LPL polymorphism (D9N) predicts cardiovascular disease risk directly and through interaction with CETP polymorphism (TaqIB) in women with high HDL cholesterol and CRP. Atherosclerosis 2011;214:373-6. [23] Corsetti JP, Ryan D, Rainwater DL, et al. CETP polymorphism (TaqIB) associates with risk in postinfarction patients with high HDL cholesterol and high CRP levels. Arterioscler Thromb Vasc Biol 2010;30:1657-64. [24] Corsetti JP, Ryan D, Moss AJ, et al. NAD(P)H oxidase polymorphism (C242T) and high HDL-C associate with recurrent coronary events in post infarction patients. Atherosclerosis 2007;196:461-8. [25] Huang Y. Mechanisms linking apolipoprotein E isoforms with cardiovascular and neurological diseases. Curr Opin Lipidol 2010;21:337-45.
[26] Greenow K, Pearce NJ, Ramji DP. The key role of apolipoprotein E in atherosclerosis. J Mol Med 2005;83:329-42. [27] Jofre-Monseny L, Minihane AM, Rimbach G. Impact of apoE genotype on oxidative stress, inflammation and disease risk. Mol Nutr Food Res 2008;52:131-45. [28] Pinto-Sietsma SJ, Janssen WMT, Hillege HL, et al. Urinary albumin excretion is associated with renal functional abnormalities in a non-diabetic population. J Am Soc Nephrol 2000;11:1882-8. [29] Lambers Heerspink HJ, Brantsma AH, de Zeeuw D, et al. Albuminuria assessed from first-morning-void urine samples versus 24 hour urine collections as a predictor of cardiovascular morbidity and mortality. Am J Epidemiol 2008;168: 897-905. [30] Verhave JD, Gansevoort RT, Hillege HL, et al. An elevated urinary albumin excretion predicts de novo development of renal function impairment in the general population. Kidney Int 2004;66:S18-21. [31] Borggreve SE, Hillege HL, Wolffenbuttel BHR, et al. An increased coronary risk is paradoxically associated with common cholesteryl ester transfer protein gene variations that relate to higher high-density lipoprotein cholesterol: a population-based study. J Clin Endocrinol Metab 2006;91: 3382-8. [32] Corsetti JP, Zareba W, Moss AJ, et al. Serum glucose and triglyceride determine high-risk subgroups in nondiabetic postinfarction patients. Atherosclerosis 2005;183: 293-300. [33] Mooijaart SP, Berbee JFP, van Heemst D, et al. ApoE plasma levels and risk of cardiovascular mortality in old age. PLoS Med 2006;3:874-83. [34] Reilly M, Rader DJ. Apolipoprotein E and coronary disease: a puzzling paradox. PLoS Med 2006;3:736-8. [35] Soderlund S, Watanabe H, Ehnholm C, Jauhiainen M, et al. Increased apolipoprotein E level and reduced high-density lipoprotein mean particle size associate with low highdensity lipoprotein cholesterol and features of metabolic syndrome. Metab Clin Exp 2010;60:1502-9. [36] Mahley RW, Weisgraber KH, Huang Y. Apolipoprotein E: structure determines function, from atherosclerosis to Alzheimer’s disease to AIDS. J Lipid Res 2009;50: S183-8. [37] Huang Y, Liu XQ, Rall SC, et al. Overexpression and accumulation of apolipoprotein E as a cause of hypertriglyceridemia. J Biol Chem 1998;273:26388-93.