Lipoprotein(a) in American Indians is Low and Not Independently Associated with Cardiovascular Disease

Lipoprotein(a) in American Indians is Low and Not Independently Associated with Cardiovascular Disease: The Strong Heart Study WENYU WANG, PHD, DONGSHENG HU, MD, MPH, ELISA T. LEE, PHD, RICHARD R. FABSITZ, MA, THOMAS K. WELTY, MD, MPH, DAVID C. ROBBINS, MD, J.L.YEH, PHD, AND BARBARA V. HOWARD, PHD

PURPOSE: To evaluate the distribution of lipoprotein(a) (Lp(a)) and assess its association to cardiovascular disease (CVD) in American Indians. METHODS: Lp(a) was measured in 3991 American Indians (aged 45–74 years with no prior history of CVD at baseline) from 13 communities in Arizona, Oklahoma, and South/North Dakota. They were followed prospectively from 1989 to 1997 for CVD. The distribution of Lp(a) was examined by center, sex, and diabetic status. Spearman correlation coefficients and Cox regression models were used to evaluate the association of Lp(a) to CVD. RESULTS: A total of 388 participants subsequently developed CVD. Median Lp(a) concentration in American Indians was 3.0 mg/dl. This was almost half of that in whites and one sixth in blacks from the CARDIA study measured by the same method. Nondiabetic participants had significantly higher Lp(a) levels than diabetic participants for both genders. Lp(a) levels were higher in women than in men for nondiabetic participants, but there was no gender difference for diabetic participants. Correlation analysis showed Lp(a) was significantly negatively correlated with the degree of Indian heritage, insulin, triglycerides (TG), fasting plasma glucose (FPG), and 2-hour plasma glucose (2hPG), and positively with low-density lipoproteins (LDL), apoprotein B (apoB), and fibrinogen (FIB). In Cox regression models, adjusting for other risk factors, Lp(a) was no longer a significant predictor of CVD in either diabetic or nondiabetic participants. CONCLUSIONS: The lower concentration of Lp(a) in American Indians and the high correlation with Indian heritage confirm the concept that Lp(a) concentration is in large part genetically determined. Lp(a) concentration is not an independent predictor of CVD among American Indians; it is higher in those who develop CVD because of its positive correlation with LDL, apoB, and FIB. Ann Epidemiol 2002;12:107–114. © 2002 Elsevier Science Inc. All rights reserved. KEY WORDS:

Lipoprotein(a), American Indians, Cardiovascular disease, Cox regression model, the Strong

Heart Study.

INTRODUCTION Since lipoprotein (a) (Lp(a)) was first reported by Berg in 1963 (1) as a sinking pre-beta lipoprotein band that represents an antigenically distinct component of the low-density lipoproteins (LDL) fraction, there has been intense investigations of its structure and function. The structure of Lp(a) is reasonably well understood. It is a complex of LDL From the Center for American Indian Health Research, College of Public Health, University of Oklahoma, Oklahoma City, OK (W.W., E.T.L., J.L.Y.); College of Public Health, Henan Medical University, Zhengzhou, China (D.H.); MedStar Research Institute, Washington, DC (D.H.); The National Heart, Lung and Blood Institute, NIH, Bethesda, MD (R.R.F.); Aberdeen Area Tribal Chairmen’s Health Board, Aberdeen, SD (T.K.W.); and MedStar Research Institute, Washington, DC (D.C.R., B.V.H.). Address reprint requests to: Wenyu Wang, Ph.D., Center for American Indian Health Research, College of Public Health, OUHSC, P.O. Box 26901, Oklahoma City, OK 73190. This manuscript does not necessarily represent the views of the Indian Health Service. Received February 19, 2001; revised May 9, 2001; accepted June 28, 2001. © 2002 Elsevier Science Inc. All rights reserved. 655 Avenue of the Americas, New York, NY 10010

to which an additional protein component, apolipoprotein (a) (apo(a)), is linked via a disulfide bridge to apoprotein B (apoB) (2). This special structure gives Lp(a) thrombotic properties in addition to atherogenic capacity (3, 4). Lp(a) is considered to be a biomarker associated with increased risk for cardiovascular disease (CVD) and atherosclerotic disease (5, 6). Some longitudinal epidemiological studies in different populations support this hypothesis (7– 12). A meta analysis indicated that in 12 of 14 prospective studies, Lp(a) concentrations were higher in subjects who later developed ischemic heart disease than in those who did not, and the extent of the difference was similar in men and women. These findings provide evidence in support of a causal role for Lp(a) in the development of atherosclerosis (13). This relationship between Lp (a) and CVD also exists in cross sectional analyses of patients with diabetes, dyslipidemia, and hemodialysis (14–20). The enhanced accumulation of Lp(a) in the arterial wall (21), or its action to inhibit fibrinolysis have been proposed to explain the increased risk of CVD (22, 23). 1047-2797/02/$–see front matter PII S1047-2797(01)00273-3

108

Wang et al. Lp(a) AND CVD IN AMERICAN INDIANS

Selected Abbreviations and Acronyms apo(a)  apolipoprotein (a) CVD  cardiovascular disease apoB  apoprotein B (apoB) UACR  ratio of albumin (mg) to creatinine (g) FPG  fasting plasma glucose FIB  fibrinogen HbA1c  glycosylated hemoglobin HDL  high-density lipoprotein Lp(a)  lipoprotein(a) LDL  low-density lipoproteins SHS  the Strong Heart Study TG  triglycerides 2hPG  2-hour plasma glucose

On the other hand, Lp(a) is found to be highly related to other CVD risk factors such as cholesterol, LDL, triglycerides (TG), and fibrinogen (FIB). These relationships may represent part of the mechanism of the association of Lp(a) to CVD. Several of the longitudinal studies examining the relationship between Lp(a) and CVD did not adjust for these possible-confounding variables (24–26). Moreover, some studies have shown that Lp(a) is not an independent risk factor for CVD (27–30). These contradictory findings indicate that more data based on prospective studies with large sample size are needed to resolve the issue. More data in diverse ethnic groups are also needed since Lp(a) concentration varies widely by ethnicity (24, 31). In this paper, we use Lp(a) as well as the other CVD risk factors data measured at the baseline examination for 4549 American Indian participants in the Strong Heart Study (SHS), and the CVD mortality and morbidity follow up surveillance data observed from the baseline examination through 1997. The SHS is a longitudinal and population based study of CVD and its risk factors in American Indians (32–34). The data provide unique opportunity to examine the association of Lp(a) concentration to CVD in American Indians.

STUDY POPULATION AND METHODS Data were analyzed from 3991 out of 4549 SHS baseline exam participants, who had no prior history of CVD and whose Lp(a) values were available. Between the baseline examination and the end of 1997, a total of 388 participants subsequently developed CVD (134 fatal and 254 nonfatal CVD cases). The baseline exam participants were 45 to 74 year old American Indians from 13 communities in Arizona, Oklahoma, and South/North Dakota. The details of the study design, survey methods, and laboratory techniques have been reported previously (32–34). The examinations for the study consisted of a personal interview, a physical examination including a 12-lead resting electrocardiogram, and laboratory tests (32–34). Participants were

AEP Vol. 12, No. 2 February 2002: 107–114

asked to bring their medications with them and were examined in the morning after at least a 12-hour overnight fast. Fasting blood samples were drawn for various measurements including glucose, lipids, lipoproteins, insulin, plasma creatinine, plasma fibrinogen and HbA1c. Informed consent was obtained. A 75-g oral glucose tolerance test was also performed as described previously (35). Lp(a) (total mass) was measured as described previously (24). Percentage of Indian heritage was computed from reported degree of Indian heritage for each parent and grandparent. Participants were classified as diabetic according to American Diabetes Association criteria (36). Type 1 diabetes is very rare in American Indians. Therefore, all diabetes in the SHS is considered to be type 2 diabetes. Urinary albumin excretion was estimated by the ratio of albumin (mg) to creatinine (g) (UACR). Deaths among the original SHS cohort between the baseline examination and the end of 1997 were identified through tribal and Indian Health Service (IHS) hospital records and by direct contact with participants and their families by study personnel. Copies of death certificates were obtained from state health departments and International Classificaiton of Diseases (ICD-9) coded centrally by a nosologist. A possible CVD death is defined as mention of any of the following anywhere on the death certificate: any type of CVD (ICD9 codes 390–448), diabetes (ICD9 code 250), diseases of the lung (ICD9 code 518.4), renal disease (ICD9 code 585), and sudden death with unknown cause (ICD9 code 789). The details are described previously (32, 37). Cause of death was investigated through autopsy reports, medical record abstractions, and informant interviews, as described previously (37). All materials were reviewed independently by physician members of the SHS Mortality Review Committee to confirm the cause of death. Criteria for fatal CVD were as described previously (32, 37). Medical records were reviewed to identify any nonfatal CVD events that had occurred since the baseline examination. New myocardial infarction (MI) and new CVD events were defined as in the baseline examination (38). For all potential CVD events or interventions, medical records were reviewed by trained medical record abstractors. Records of outpatient visits were reviewed and abstracted for procedures diagnostic of CVD (e.g., treadmill tests, coronary angiography). Information obtained from the chart review was reviewed by a physician member of the SHS Mortality or Morbidity Review Committee in order to establish the specific CVD diagnosis (32). Blinded review of the abstracted records by other physician members of the Morbidity Review Committee showed  90% concordance in diagnosis. Statistical Analysis Stratified analyses by diabetic status are provided beside the overall analyses in this study since American Indians have a very high diabetes mellitus (DM) prevalence (1937 out of

AEP Vol. 12, No. 2 February 2002: 107–114

109

Wang et al. Lp(a) AND CVD IN AMERICAN INDIANS

3991 were diabetic participants), DM is a very strong determinant of CVD, and most of the CVD cases (262 out of 388) were diabetic participants. Because Lp(a) levels were highly skewed, median and range are presented by center and gender, or by DM status and gender, or by CVD status and DM status. Accordingly, the median differences among different groups were assessed by Wilcoxon’s rank-sum test or its expanded version, the Kruskal-Wallis test. Spearman correlation coefficients by DM status were used to examine the relationships between Lp(a) and DM measurements as well as some selected CVD risk factors. Adjusted relative CVD risks (relative hazard ratios) of the percentile-divided Lp(a) levels: (25th, 50th), (50th, 75th), (75th, 90th), and  90th, to  25th (referent) were obtained by applying a Cox proportional hazards model and adjusting age, gender, Indian heritage, BMI, center, smoking status, SBP, FPG, and HDL for all participants or by DM status. The relative CVD risks were adjusted further by adding TG, UACR, LDL, FIB and apoB separately to the basic adjusted models to assess the possible joint effects of Lp(a) and these additional measures. Testing of the trend of increasing levels of Lp(a) to the risk of CVD was also performed in the adjusted models. The disease-free time for a participant in the Cox proportional hazards model was defined as the duration from the baseline examination to the time of the first CVD event or the cutoff time (December 31, 1997) or last-follow-up before the cutoff time. Moreover, the adjusted relative CVD risks (relative hazard ratios) for one unit increase of log-transformed Lp(a) concentration, Log(Lp(a)), was also derived by using a Cox proportional hazards model adjusting for all previously identified covariates. P-values  0.05 were considered to be statistically significant.

dian, mean and standard deviation of Lp(a) levels for white and black participants in the CARDIA study (24) by gender. Menopausal status and estrogen usage were not considered in this study since most of the women (77.3%; 1852 out of 2396 women) were postmenopausal and only small portion (11%) of the postmenopausal women were taking estrogen (39). In fact, the Lp(a) levels in postmenopausal women were not significantly higher than those in nonpostmenopausal women (both groups had the same median 3.0, p  0.054). There were no differences in Lp(a) levels between males and females within each of the three centers, but there were significant differences in Lp(a) levels among the three SHS centers for male (p  0.0001) and female (p  0.0001) participants. Arizona had the lowest median, which was followed by South/North Dakota. Oklahoma American Indians had the highest median Lp(a). White and blacks in the CARDIA had considerably higher (twice and six-times, respectively) Lp(a) medians than those of the SHS American Indian participants in all three centers. The Lp(a) levels in diabetic participants were significantly lower than those in nondiabetic participants (p  0.0001) and for either men (p  0.0001) or women (p  0.0001) or for all participants (p  0.0001) (Table 2). Women had higher Lp(a) than men in nondiabetic participants (p  0.02) but not in diabetic participants (p  0.67) (Table 2). There were significant differences in the baseline Lp(a) levels between those who developed CVD and those who did not in diabetic participants (p  0.01), but this was not the case in nondiabetic participants (p  0.06) (Table 2). The Spearman correlation coefficients between Lp(a) and DM measurements by DM status (Table 3) showed significant negative correlations of Lp(a) with FPG (p  0.0001), 2hPG (p  0.0006), insulin (p  0.0001), and UACR (p  0.0001) in nondiabetic participants, and only with FPG (p  0.026), 2hPG (p  0.0001), and insulin (p 

RESULTS Table 1 displays the median and range of Lp(a) levels for American Indian participants in three centers, and the me-

TABLE 1. Lp(a) (mg/dl) for American Indians in SHS American Indians in SHS Arizona

Oklahoma

CARDIA study S/N Dakota

Black

White

Percentile

Men

Women

Men

Women

Men

Women

Men

Women

Men

Women

95th 90th 75th 50th 25th 10th 5th N

10.0 7.0 4.0 2.0 1.0 1.0 1.0 482

11.3 8.0 4.0 2.0 1.0 1.0 0.8 844

39.0 24.0 9.0 4.2 1.9 1.0 1.0 549

38.1 23.0 8.6 4.0 2.0 1.0 1.0 781

25.0 15.0 7.0 3.0 1.3 1.0 1.0 564

29.0 14.0 7.0 3.4 2.0 1.0 1.0 771

67.9 58.1 37.9 21.5 9.8 4.4 2.0 861

72.9 62.1 42.8 23.9 12.1 5.5 2.9 1128

52.7 39.1 19.8 6.1 2.2 1.0 0.6 1011

51.5 39.9 20.2 6.4 2.8 1.3 0.9 1125

Lp(a) for both studies were measured in the same laboratory; within men or women, there are median differences among the three SHS centers ( p  0.0001); within each SHS center, there are no median differences between men and women (p  0.09).

110

Wang et al. Lp(a) AND CVD IN AMERICAN INDIANS

AEP Vol. 12, No. 2 February 2002: 107–114

TABLE 2. Lp(a) level by gender, diabetic, and CVD Status All Men Women P

All Nondiabetic Diabetic P

Nondiabetic

Diabetic

P

4.0 (0.19–97.6) (N  2054) 3.3 (0.57–85.3) (N  900) 4.0 (0.19–97.6) (N  1154) 0.02

2.5 (0.19–96.3) (N  1937) 2.1 (0.57–78.0) (N  695) 2.8 (0.19–96.3) (N  1242) 0.67

0.0001

NonCVD

CVD

P

3.0 (0.19–97.6) (N  3603) 4.0 (0.19–97.6) (N  1928) 2.3 (0.19–96.3) (N  1675) 0.0001

3.4 (0.19–85.3) (N  388) 4.7 (0.57–85.3) (N  126) 3.0 (0.19–80.8) (N  262) 0.002

0.08

with Lp(a) in either diabetic or nondiabetic participants (Table 3). Moreover, the proportion of the variation of Lp(a) level explained by each variable (Table 3), represented by 100*R2 from the regression model of Log(Lp(a)) on each respective variable, indicated that LDL, ApoB, and Indian heritage carried more proportion of the variation than the other variables. Cox proportional hazards models were used to assess the association of Lp(a) levels to CVD. Percentile effects of Lp(a) were shown in Tables 4 and 5. Adjusted relative hazard ratios of CVD among participants with Lp(a) level  90th percentile ( 13 mg/dl) were significantly (1.64 times) higher than those of participants with Lp(a) level  25th percentile ( 1.15 mg/dl) after adjusting for age, gender, Indian heritage, BMI, center, smoking status, SBP, FPG, and HDL (Table 4). After further adjustment by adding UACR, LDL, FIB, or apoB individually to the basic variables, the relative hazard ratios were further reduced (Table 4). When the same analyses were performed for diabetic and nondiabetic participants separately, the significant percentile effect was found in the first model, but it was diminished after adding UACR, LDL, FIB, or apoB in diabetic participants (Table 5). In nondiabetic participants, there were no such the significant percentile effect between Lp(a) and CVD in any of the models (Table 5). The continuous effect of Lp(a) was shown in Table 6. For one unit increase of Log(Lp(a)), the adjusted relative hazard ratio of CVD was 1.14 and 1.18 times for diabetic participants in the models adjusted by the basic adjusted variables or by adding TG to the basic adjusted vari-

0.0001 0.0001

0.06 0.01

Data are presented as median (mg/dl) (range).

0.0001) in diabetic participants. HbA1c was not related to Lp(a) in either diabetic (p  0.8) or non-diabetic (p  0.6) participants (Table 3). Lp(a) was correlated positively with LDL (p  0.0001), HDL (p  0.03), apoA1 (p  0.006), apoB (p  0.0001), and FIB (p  0.0009), and negatively to BMI (p  0.001), Indian heritage (p  0.0001), and TG (p  0.006) in nondiabetic participants (Table 3). It was correlated positively with age (p  0.0001), LDL (p  0.0001), apoB (p  0.0001), and FIB (p  0.0001), and negatively to BMI (p  0.032), Indian heritage (p  0.0001) and TG (p  0.0001) in diabetic participants (Table 3). SBP was not correlated

TABLE 3. Spearman correlation coefficients between Lp(a) and indices of diabetes and selected CVD risk factors in diabetic and nondiabetic participants All (N  3991)

Fasting glucose (mg/dl) 2-hr glucose (mg/dl) HbA1c (%) Insulin (U/ml) Albumin/creatinine (mg/g) Age (years) BMI (kg/m2) SBP (mmHg) Indian heritage (%) LDL-cholesterol (mg/dl) HDL-cholesterol (mg/dl) Triglycrides (mg/dl) ApoAI (g/l) ApoB (g/l) LDL-size (Å) Fibrinogen (mg/dl) a

Nondiabetic (N  2054)

Diabetic (N  1937) a

Correla. coeff.

P

Correla. coeff.

P

(%)

0.172 0.207 0.129 0.184 0.118 0.045 0.096 0.045 0.201 0.238 0.071 0.124 0.052 0.156 0.105 0.043

 0.0001  0.0001  0.0001  0.0001  0.0001 0.0045  0.0001 0.0047  0.0001  0.0001  0.0001  0.0001 0.0010  0.0001  0.0001 0.0070

0.086 0.076 0.012 0.140 0.085 0.033 0.071 0.039 0.199 0.216 0.047 0.060 0.060 0.178 0.060 0.074

0.0001 0.0006 0.6009  0.0001 0.0001 0.1375 0.0013 0.0748  0.0001  0.0001 0.0321 0.0063 0.0063  0.0001 0.0065 0.0009

0.73 0.58 0.04 1.73 0.53 0.12 0.44 0.09 4.92 5.01 0.12 0.63 0.32 3.71 0.31 0.31

Correla. coeff.

P

(%)a

0.050 0.119 0.004 0.101 0.018 0.099 0.049 0.014 0.145 0.231 0.034 0.117 0.013 0.151 0.088 0.106

0.0268  0.0001 0.8668  0.0001 0.4159  0.0001 0.0323 0.5482  0.0001  0.0001 0.1349  0.0001 0.5733  0.0001 0.0001  0.0001

0.21 1.18 0.01 0.80 0.07 0.77 0.14 0.04 3.11 5.64 0.15 1.62 0.05 2.77 0.79 0.64

100*R2 from the regression model of Log(Lp(a)) on the respective variable; Log(Insulin), Log(Triglycerides), and Log(Albumin/creatinine) are used in the respective regression model.

AEP Vol. 12, No. 2 February 2002: 107–114

Wang et al. Lp(a) AND CVD IN AMERICAN INDIANS

111

TABLE 4. Adjusted relative risks of CVD from 25th (referent) to other percentiles of Lp(a) level based on Cox regression models for all participants (N  3991) Percentile Range(mg/dl)

25th (0, 1.15]

50th (1.15, 3]

75th (3, 6.51]

90th (6.51, 13]

 90th (13, 97.59]

Relative risk (95% CI)a Relative risk (95% CI)b Relative risk (95% CI)c Relative risk (95% CI)d Relative risk (95% CI)e Relative risk (95% CI)f

1

1.05 0.78–1.41 1.08 0.81–1.45 1.00 0.74–1.34 1.03 0.77–1.38 1.03 0.76–1.37 1.03 0.77–1.38

1.10 0.81–1.48 1.14 0.84–1.55 1.05 0.78–1.42 1.04 0.77–1.41 0.98 0.72–1.34 1.06 0.79–1.44

1.27 0.91–1.79 1.32 0.94–1.86 1.13 0.80–1.58 1.16 0.83–1.64 1.17 0.83–1.64 1.18 0.84–1.66

1.64 1.15–2.36 1.69 1.17–2.43 1.49 1.04–2.14 1.46 1.01–2.11 1.51 1.05–2.17 1.51 1.05–2.17

1 1 1 1 1

P for trend 0.038 0.023 0.158 0.155 0.156 0.114

a

Adjusted by age, gender, BMI, center, smoking, SBP, Indian heritage, FPG, and HDL. b Adjusted by variables in a and triglycerides. c Adjusted by variables in a and albumin/creatinine. d Adjusted by variables in a and LDL. e Adjusted by variables in a and fibrinogen. f Adjusted by variables in a and apoB.

ables, respectively (Table 6). However, the hazard ratios were no longer significant for diabetic participants in the other models adjusted by adding UACR, LDL, FIB, or apoB separately to the basic adjusted variables (Table 6). Similar modeling patterns were true for all participants (Table 6). However, hazard ratios were never significant in all models for nondiabetic participants (Table 6).

DISCUSSION The SHS data have provided an opportunity for a prospective and population-based analysis of Lp(a) concentrations in American Indians and assessment of associations of Lp(a) to CVD adjusted for other CVD risk factors. American Indians have low concentrations of Lp(a). Measured in the same laboratory, the medians of Lp(a) in American Indians were 3 to 1.5 times lower than those of whites (about 6.1–6.4 mg/dl), and 10 to 5 times lower than those of blacks (about 21.5–24 mg/dl) in the CARDIA study (24). The 90th percentile (mg/dl) of Lp(a) in American Indians was 13 mg/dl, which was three times lower than those in whites (about 40) and 4–5 times lower than in blacks (about 60) (24). There was only a small percentage of American Indians with Lp(a)  30 mg/dl (considered to be an elevated or a risk level in the literature (40,41), 1.73%, 7.37%, and 4.34% in Arizona, Oklahoma, and South/North Dakota, respectively). Despite the low concentration of Lp(a) in American Indians, there were significant population differences. The median of Lp(a) concentration in Oklahoma American Indians was two times higher than those in Arizona for either

gender. Moreover, Lp(a) was also significantly and negatively related to the degree of Indian heritage. Oklahoma Indians report a greater proportion of non-Indian admixture than do the other two centers. These ethnic and center differences as well as the high correlation with Indian heritage confirm further the general consensus in the literature that Lp(a) concentrations are in large part genetically determined (24, 42–44). The results showed that American Indian diabetic participants had significantly lower Lp(a) levels than nondiabetic participants (median 2.5 vs. 4.0 mg/dl; 2.1 vs. 3.3 mg/ dl in men; 2.8 vs. 4.0 in women; p  0.0001; Table 2). Lp(a) levels have been reported to be high (45–47) or unchanged (48,49) in individuals with type 2 DM. It is possible that the different findings in participants with diabetes are a reflection of genetic heritage since Indian ancestry is associated with increasing diabetes rates. Thus, diabetes in other populations may be associated with ethnic groups who have higher or lower Lp(a) levels. LDL was the most significant positive correlate of Lp(a) (r  0.216 and 0.231 for nondiabetic and diabetic participants, respectively), then ApoB (r  0.178 and 0.151), and FIB (r  0.074 and 0.106) in American Indians (Table 3). These findings were similar to those observed in whites and blacks (24). There were also small but significant negative correlations of Lp(a) to insulin (r  0.14 and 0.101 for nondiabetic and diabetic participants, respectively), TG (r  0.06 and 0.117), 2hPG (r  0.076 and 0.119) and FPG (r  0.086 and 0.05) in American Indians (Table 3). These significant negative correlations were not observed in whites and blacks (24). These differences might be due to the very high DM prevalence in American Indi-

112

Wang et al. Lp(a) AND CVD IN AMERICAN INDIANS

AEP Vol. 12, No. 2 February 2002: 107–114

TABLE 5. Adjusted relative risks of CVD from 25th (referent) to other percentiles of Lp(a) level based on Cox regression models for diabetic or nondiabetic participants Percentile Range(mg/dl)

25th

50th

75th

90th

90th

(0, 1.15]

(1.15, 3]

(3, 6.51]

(6.51, 13]

(13, 97.59]

P for trend

Diabetic (N  1937) Relative risk (95% CI)a Relative risk (95% CI)b Relative risk (95% CI)c Relative risk (95% CI)d Relative risk (95% CI)e Relative risk (95% CI)f

1

Relative risk (95% CI)a Relative risk (95% CI)b Relative risk (95% CI)c Relative risk (95% CI)d Relative risk (95% CI)e Relative risk (95% CI)f

1

1 1 1 1 1

1 1 1 1 1

1.07 0.76–1.51 1.15 0.81–1.62 0.98 0.69–1.39 1.05 0.75–1.48 1.06 0.75–1.50 1.03 0.73–1.46

1.08 1.4 0.75–1.55 0.92–2.11 1.19 1.52 0.82–1.73 1.00–2.31 1.01 1.16 0.69–1.46 0.77–1.76 1.01 1.19 0.70–1.46 0.79–1.81 0.98 1.25 0.68–1.41 0.83–1.89 1.05 1.2 0.73–1.51 0.79–1.82 Nondiabetic (N  2054)

1.84 1.17–2.88 2.05 1.30–3.25 1.58 1.00–2.49 1.57 0.99–2.48 1.66 1.05–2.62 1.6 1.02–2.53

1.03 0.58–1.82 1 0.57–1.77 1.03 0.58–1.83 1.02 0.58–1.80 0.99 0.56–1.75 1.03 0.58–1.82

1.17 0.67–2.05 1.13 0.64–1.97 1.21 0.69–2.12 1.13 0.65–1.98 1.08 0.61–1.90 1.17 0.67–2.05

1.58 0.85–2.95 1.45 0.77–2.72 1.58 0.84–2.97 1.45 0.77–2.74 1.5 0.80–2.79 1.57 0.84–2.94

1.2 0.64–2.22 1.15 0.62–2.14 1.18 0.63–2.20 1.15 0.62–2.14 1.13 0.61–2.10 1.19 0.64–2.21

0.030 0.007 0.200 0.170 0.120 0.140

0.320 0.440 0.320 0.430 0.450 0.330

a

Adjusted by age, gender, BMI, center, smoking, SBP, Indian heritage, FPG, and HDL. b Adjusted by variables in a and triglycerides. c Adjusted by variables in a and albumin/creatinine. d Adjusted by variables in a and LDL. e Adjusted by variables in a and fibrinogen. f Adjusted by variables in a and apoB.

ans (35). Relations between Lp(a) and glucose and insulin were also observed in a Mexican Americans population another population with high prevalence of DM (50). The negative correlation between Lp(a) and TG in American Indians was also reported in other populations such as the Framingham study (51), a hyperlipidemic Italian population (52), a group with CVD (53), and a group with diabetes (45). The cause of this inverse relation is unknown (45,54,55). Although most of the variables shown in Table 3 have significant (p  0.05) correlations with Lp(a), the proportions of the variation of Lp(a) level carried by even the most highly correlated variables (LDL, ApoB, and Indian heritage) was less than 6% (Table 3). This is consistent with most studies in the literature (e.g., 24, 44) and suggests strong heritability. Cox proportional hazards models explored the association of Lp(a) with CVD. The analysis of the relation between Lp(a) and CVD was dependent on LDL, apoB, and FIB, which had significant linear correlations with Lp(a).

Association of Lp(a) to CVD was diminished or no longer significant when LDL, apoB, and FIB were added to the Cox proportional hazards models. The largest influence was with LDL, which is a well-established risk factor of CVD. To consider a possible threshold effect, analyses were performed in which a dichotomous variable defined by the 90th percentile cut-point of Lp(a) was used. The results were similar to those shown in Tables 4 and 5 (data not shown). Among all 396 participants with Lp(a)  the 90th percentile cut-point, only 53 (13.38%) developed CVD. Thus, conflicting reports in the literature concerning Lp(a) as a predictor of CVD may be due in part to the fact that some of the studies did not adjust for CVD risk factors such as LDL, apoB, and FIB. In conclusion, our results showed that Lp(a) levels in American Indians were considerably lower than those in whites and blacks. Lp(a) was significantly positively correlated to LDL, ApoB, and FIB, but significantly negatively correlated to insulin, 2hPG, FPG, TG, and Indian heritage in American Indians. Nondiabetic participants had signifi-

AEP Vol. 12, No. 2 February 2002: 107–114

Wang et al. Lp(a) AND CVD IN AMERICAN INDIANS

TABLE 6. Adjusted relative risks of CVD for one unit change of Log(Lp(a)) level based on Cox regression models for diabetic or nondiabetic participants All (N  3991) Model 1a 2b 3c 4d 5e 6f

Diabetic (N  1937)

Nondiabetic (N  2054)

Relative Relative Relative risk (95% CI) risk (95% CI) risk (95% CI) 1.12 1.14 1.09 1.08 1.09 1.09

1.02–1.24 1.03–1.25 0.99–1.20 0.98–1.20 0.99–1.21 0.99–1.21

1.14 1.18 1.09 1.08 1.10 1.09

1.01–1.28 1.04–1.33 0.97–1.23 0.95–1.22 0.97–1.24 0.96–1.23

1.14 1.12 1.13 1.12 1.12 1.14

0.96–1.35 0.94–1.33 0.95–1.34 0.94–1.33 0.95–1.33 0.96–1.35

12.

13.

14.

15.

a

Adjusted by age, gender, BMI, center, smoking, SBP, Indian heritage, FPG, and HDL. b Adjusted by variables in a and triglycerides. c Adjusted by variables in a and albumin/creatinine. d Adjusted by variables in a and LDL. e Adjusted by variables in a and fibrinogen. f Adjusted by variables in a and apoB.

16.

17.

18.

cantly higher Lp(a) levels than diabetic participants. Lp(a) was not an independent risk factor for CVD; initial association with CVD were diminished when LDL, apoB, and FIB were considered.

19.

20.

REFERENCES 1. Berg K. A new serum type system in man: The Lp system. Acta Pathol Microbiol Scand. 1963;59:369–382. 2. Fless GM, ZumMallen ME, Scanu AM. Physiochemical properties of apolipoprotein (a) and lipoprotein (a) derived from the dissociation of human plasma lipoprotein (a). J Biol Chem. 1986;261:8712–8718. 3. Ziegler O, Guerci B, Candiloros H, Drouin P. Lipoprotein (a) and diabetes mellitus. Diabete Metab. 1995;21(2):127–138. 4. Couderc R, Maachi M. Lipoprotein (a), a cardiovascular risk factor: Importance of its determination in current clinical practice. J Maladies Vascul. 1998;23(3):165–175. 5. Frishman WH. Biologic markers as predictors of cardiovascular disease. Am J Med. 1998;104(6A):18S–27S. 6. Oparil S, Oberman A. Nontraditional cardiovascular risk factors. Am J Med Sci. 1999;317(3):193–207. 7. Bostom AG, Gagnon DR, Cupples LA, Wilson PW, Jenner JL, Ordovas JM, et al. A prospective investigation of elevated lipoprotein (a) detected by electrophoresis and cardiovascular disease in women. The Framingham Heart Study. Circulation. 1994;90(4):1688–1695. 8. Berg K, Dahlen G, Christophersen B, Cook T, Kjekshus J, Pedersen T. Lp(a) lipoprotein level predicts survival and major coronary events in the Scandinavian Smivastatin Survival Study. Clin Genet. 1997; 52(5):254–261. 9. Dahlen GH, Stenlund H. Lp(a) lipoprotein is a major risk factor for cardiovascular disease: Pathogenic mechanisms and clinical significance. Clin Genet. 1997;52(5):272–280. 10. Assmann G, Schulte H, Eckardstein A. Hypertriglyceridemia and elevated lipoprotein (a) are risk factors for major coronary events in middle-aged men. Am J Cardiol. 1996;77:1179–1184. 11. Nguyen TT, Ellefson RD, Hodge DO, Bailey KR, Kottke TE, AbuLebdeh HS. Predictive value of electrophoretically detected lipopro-

21.

22.

23.

24.

25. 26.

27.

28.

29.

113

tein (a) for coronary heart disease and cerebrovascular disease in a community-based cohort of 9936 men and women. Circulation. 1997;96(5):1390–1397. Catalano M, Perilli E, Caizaniga G, Colombo F, Carotta M, Andreoni S. Lp(a) in hypertensive patients. J Human Hypertens. 1998; 12(2):83–89. Craig WY, Neveux LM, Palomaki GE, Cleveland MM, Haddow JE. Lipoprotein (a) as a risk factor for ischemic heart disease: Metaanalysis of prospective studies. Clin Chem. 1998;44(11):2301–2306. Heesen BJ, Wolffenbutte BH, Leurs PB, Sels JP, Menheere PP, JackleBeckers SE, et al. Lipoprotein (a) levels in relation to diabetic complications in patients with non-insulin-dependent diabetes. Eur J Clin Invest. 1993;23(9):580–584. Ruiz J, Thillet J, Huby T, James RW, Erlich D, Flandre P, et al. Association of elevated lipoprotein (a) levels and coronary heart disease in NIDDM patients. Relationship with apolipoprotein (a) phenotypes. Diabetologia. 1994;37(6):585–591. Miyao M, Araki A, Hattori A, Miyachi T, Inoue J, Horiuchi T, et al. Lipoprotein (a) as a risk factor for cardiovascular disease in elderly patients with diabetes. Jpn J Geriatr. 1997;34(3):185–191. Mohan V, Deepa R, Haranath SP, Premalatha G, Rema M, Sastry NG, et al. Lipoprotein (a) is an independent risk factor for coronary artery disease in NIDDM patients in South India. Diabetes Care. 1998;21(11):1819–1823. Arsenio L, Cioni F, Japichino GG, Strata A. Lp(a) and the cardiovascular risk in dyslipidemia subjects. Acta Bio-Med Ateneo Parmense. 1998;69(3–4):97–103. Koda Y, Nishi S, Suzuki M, Hirasawa Y. Lipoprotein (a) is a predictor for cardiovascular mortality of hemodialysis patients. Kidney Int. 1999;71:S251–S253. Ohashi H, Oda H, Ohno M, Watanabe S, Sakata S. Lipoprotein (a) as a risk factor for coronary heart disease in hemodialysis patients. Kidney Int. 1999;71:S242–S244. Kreuzer J, Lloyd MB, Bok D, Fless GM, Scanu AM, Lusis AJ, et al. Lipoprotein (a) displays increased accumulation compared with lowdensity lipoprotein in the murine arterial wall. Chem Phys Lipids. 1994;67–68:175–190. Falco C, Estelles A, Dalmau J, Espana F, Aznar J. Influence of lipoprotein (a) levels and isoforms on fibrinolytic activity-study in families with high lipoprotein (a) levels. Thromb. Haemost. 1998;79(4):818– 823. Estelles A, Cano A, Falco C, Espana F, Gilabert J, Grancha S, et al. Lipoprotein (a) levels and isoforms and fibrinolytic activity in postmenopause-influence of hormone replacement therapy. Thromb Haemost. 1999;81(1):104–110. Howard BV, Le NA, Belcher JD, Flack JM, Jacobs DR, Lewis CE, et al. Concentrations of Lp(a) in Black and White young adults: Relations to risk factors for cardiovascular disease. Ann Epidemiol. 1994;4:341–350. Rainwater DL. Lp(a) concentrates are related to plasma lipid concentrations . Atherosclerosis. 1996;127(1):13–18. Cardoso-Saldana G, Ize-Lema I, Kimura LY, Zamora Gonzalez J, Posadas Romero C. Lipoprotein (a) as cardiovascular risk in adult Mexicans. Revista Invest Clin. 1997;49(2):85–92. Ridker PM, Hennekens CH, Stampfer MJ. A prospective study of lipoprotein (a) and the risk of myocardial infarction. JAMA. 1993;270:2195–2199. O’Brien T, Nguyen TT, Harrison JM, Bailey KR, Dyck PJ, Kottke BA. Lipids and Lp(a) lipoprotein levels and coronary artery disease in subjects with non-insulin-dependent diabetes mellitus. Mayo Clin Proc. 1994;69(5):430–435. Nascetti S, D’Addato S, Pascarelli N, Sangiorgi Z, Grippo MC, Gaddi A. Cardiovascular disease and Lp(a) in the adult population and in the elderly: The Brisighella Study. Rivista Eur Sci Med Farmacol. 1996;18(5–6):205–212.

114

Wang et al. Lp(a) AND CVD IN AMERICAN INDIANS

30. Furuseth K, Berg K, Hanssen KF, Rutle O, Bruusgaard D, Vaaler S. Serum lipoprotein (a) and cardiovascular disease in non-insulin-dependent diabetes mellitus. Scand J Prim Health Care. 1998;16(1):40–43. 31. Low PS, Heng CK, Saha N, Tay JS. Racial variation of cord plasma lipoprotein (a) levels in relation to coronary risk level: A study in three ethnic groups in Singapore. Pediatr Res. 1996;40(5):718–722. 32. Lee ET, Welty TK, Fabsitz RR, Cowan LD, Le NA, Oopik AJ, et al. The Strong Heart Study. A study of cardiovascular disease in American Indians: Design and methods. Am J Epidemiol. 1990;132:1141–1155. 33. Howard BV, Welty TK, Fabsitz RR, Cowan LD, Oopik AJ, Le NA, et al. Risk factors for coronary heart disease in diabetic and nondiabetic Native Americans: The Strong Heart Study. Diabetes. 1992;41(Suppl 2):S4–S11. 34. Welty TK, Lee ET, Yeh JL, Cowan LD, Go O, Fabsitz RR, et al. Cardiovascular disease risk factors among American Indians: The Strong Heart Study. Am J of Epidemiol. 1995;142(3):269–287. 35. Lee ET, Howard BV, Savage PJ, Cowan LD, Fabsitz RR, Oopik AJ, et al. Diabetes and impaired glucose tolerance in 3 American Indian populations aged 45 to 74 years: The Strong Heart Study. Diabetes Care. 1995;18:599–610. 36. The Expert Committee on the Diagnosis and Classification of Diabetes Mellitus. Report of the Expert Committee on the diagnosis and classification of diabetes mellitus. Diabetes Care. 1997;20(7):1183–1197. 37. Lee ET, Cowan LD, Welty TK, Sievers M, Howard WJ, Oopik AJ, et al. All-cause mortality and cardiovascular disease mortality in 3 American Indian populations aged 45 to 74 years, 1984 to 88: The Strong Heart Study. Am J Epidemiol. 1998;147:995–1008. 38. Howard BV, Lee ET, Cowan LD, Fabsitz RR, Howard WJ, Oopik AJ, et al. Coronary heart disease prevalence and its relation to risk factors in American Indians: The Strong Heart Study. Am J Epidemiol. 1995; 142:254–268. 39. Cowan LD, Go O, Howard BV, Devereux RB, Pettitt DJ, Fabsitz RR, et al. Parity, postmenopausal estrogen use, and cardiovascular disease risk factors in American Indian women: The Strong Heart Study. J Womens Health. 1997;6(4):441–449. 40. Dahlen GH, Guyton JR, Attar M, Farmer JA, Gotto AM Jr. Association of levels of lipoprotein Lp(a), plasma lipids, and other lipoproteins with coronary artery disease documented by angiography. Circulation. 1986;74:758–765. 41. Bovet P, Rickenbach M, Wietlisbach V, Riesen W, Shamlaye C, Darioli R, et al. Comparison of serum lipoprotein(a) distribution and its correlates among Black and White populations. Int J Epidemiol. 1994;23(1):20–27.

AEP Vol. 12, No. 2 February 2002: 107–114

42. Fless GM, ZumMallen ME, Scanu AM. Isolation of apolipoprotein(a) from lipoprotein(a). J Lipid Res. 1985;26:1224–1229. 43. Utermann G. The mysteries of lipoprotein(a). Science. 1989;246: 904–910. 44. Boerwinkle E, Menzel HJ, Kraft HG, Utermann G. Genetics of the quantitative Lp(a) lipoprotein trait. III. Contribution of Lp(a) glycoprotein phenotypes to normal lipid variation. Hum Genet. 1989;82: 73–78. 45. Nakata H, Horita K, Eto M. Alteration of lipoprotein(a) concentration with glycemic control in non-insulin-dependent diabetic subjects without diabetic complications. Metabolism. 1993;42(10): 1323–1326. 46. Guillausseau PJ, Peynet J, Chanson P, Legrand A, Altman JJ, Poupon J, et al. Lipoprotein(a) in diabetic patients with and without chronic renal failure. Diabetes Care. 1992;15:976–979. 47. Heller FR, Jamart J, Honore P, Derue G, Novik V, Galanti L, et al. Serum lipoprotein(a) in patients with diabetes mellitus. Diabetes Care. 1993;16:819–823. 48. Haffner SM, Morales PA, Stern MP, Gruber MK. Lp(a) concentrations in NIDDM. Diabetes. 1992;41:1267–1272. 49. O’Brien T, Nguyen TT, Harrison JM, Bailey KR, Dyck PJ, Kottke BA. Lipids and Lp(a) lipoprotein levels and coronary artery disease in subjects with non-insulin-dependent diabetes mellitus. Mayo Clin Proc. 1994;69:430–435. 50. Rainwater DL, Haffner SM. Insulin and 2-hour glucose levels are inversely related to Lp(a) concentrations controlled for LPA genotype. Atheroscl Thromb Vasc Biol. 1998;18(8):1335–1341. 51. Jenner JL, Ordovas JM, Lamon-Fava S, Schaefer, MM, Wilson PWF, Castelli WP, et al. Effects for age, sex and menopausal status on plasma lipoprotein(a) levels. The Framingham Offspring Study. Circulation. 1993;87:1135–1141. 52. Werba JP, Safa O, Gianfrancheschi G, Michelagnoli S, Sirtori CR, Franceschini G. Plasma triglycerides and lipoprotein(a): Inverse relationship in a hyperlipidemic Italian population. Atherosclerosis. 1993;101:203–211. 53. Genest J, Jenner JL, McNamara JR, Ordovas JM, Silberman SR, Wilson PWF, et al. Prevalence of lipoprotein(a) excess in coronary artery disease. Am J Cardiol. 1991;67:1039–1045. 54. Boerwinkle E. Genetics of plasma lipoprotein(a) concentration. Curr Opin Lipidol. 1992;3:128–136. 55. Cohn JS, Law CHWJ, Sullivan DR, Hensley WJ. Plasma lipoprotein distribution of apolipoprotein(a) in the fed and fasted states. Atherosclerosis. 1991;90:59–66.