Author + information
- Received December 28, 2007
- Revision received March 2, 2008
- Accepted April 2, 2008
- Published online July 8, 2008.
- Jacqueline Suk Danik, MD, MPH⁎,†,‡,⁎ (, )
- Nader Rifai, PhD∥,
- Julie E. Buring, ScD†,‡ and
- Paul M. Ridker, MD, MPH⁎,†,‡,§
- ↵⁎Reprint requests and correspondence:
Dr. Jacqueline Suk Danik, Center for Cardiovascular Disease Prevention, Brigham and Women's Hospital, 900 Commonwealth Avenue, Boston, Massachusetts 02215.
Objectives This study assesses whether the relationship of lipoprotein(a) [Lp(a)] with cardiovascular risk may be modified by concurrent hormone replacement therapy (HT).
Background Prior studies indicate that HT decreases plasma levels of Lp(a), but few have been powered to assess whether it modifies the relationship of Lp(a) with cardiovascular disease (CVD).
Methods Lipoprotein(a) at baseline was measured among 27,736 initially healthy women, of whom 12,075 indicated active HT use at the time of blood draw at study initiation and 15,661 did not. The risk of first-ever major cardiovascular event (nonfatal myocardial infarction, nonfatal cerebrovascular event, coronary revascularization, or cardiovascular death) over a 10-year period was assessed with Cox proportional hazard models according to Lp(a) levels and HT status and adjusted for potential confounding variables.
Results As anticipated, Lp(a) values were lower among women taking HT (median 9.4 mg/dl vs. 11.6 mg/dl, p < 0.0001). In women not taking HT, the hazard ratio of future CVD for the highest Lp(a) quintile compared with the lowest was 1.8 (p trend <0.0001), after adjusting for age, smoking, blood pressure, diabetes, body mass index, total cholesterol, high-density lipoprotein, C-reactive protein, and treatment arms of aspirin and vitamin E. In contrast, among women taking HT, there was little evidence of association with CVD (hazard ratio: 1.1, p trend = 0.18; interaction p value = 0.0009 between Lp(a) quintiles and HT on incident CVD).
Conclusions The relationship of high Lp(a) levels with increased CVD is modified by HT. These data suggest that the predictive utility of Lp(a) is markedly attenuated among women taking HT and may inform clinicians' interpretation of Lp(a) values in such patients. (Women's Health Study [WHS]; NCT00000479)
Lipoprotein(a) [Lp(a)] is a lipoprotein that differs from low-density lipoprotein cholesterol (LDL-C) by the presence of an apolipoprotein(a) [apo(a)] component that has marked size heterogeneity (1). It has been hypothesized that its structural homology to plasminogen (2) may result in its contribution to a thrombogenic milieu (3,4). Using an assay independent of apo(a) isoform size (5,6), we have recently shown that women with high levels of Lp(a) have increased incidence of cardiovascular disease (CVD), particularly if they also have high LDL-C levels (7), in concordance with work by other groups (8–14). However, data about what modifies the relationship of Lp(a) with CVD are scarce. In terms of lifestyle, changes in diet and exercise are not known to decrease Lp(a) levels (15). In terms of medications, statin therapy may not decrease Lp(a) levels, but by lowering LDL-C, may modify the cardiovascular risk associated with Lp(a) (16); niacin, although difficult to tolerate in doses needed to decrease Lp(a) (17–19), does decrease Lp(a) levels. The data about the relationship between hormone replacement therapy (HT) and Lp(a) levels are conflicting and controversial, in part because HT use in women is itself controversial (20–25) and because studies that show decreases in Lp(a) levels with HT use (26–31) have rarely been powered to assess for impact on CVD (32).
To explore this issue in detail, we undertook a study in which we assessed whether the relationship of Lp(a) levels with CVD is modified by concurrent use of HT among 27,736 women.
Lipoprotein(a) was measured among 12,075 women taking HT and 15,661 not taking HT at the initiation of the study, and assessed for future cardiovascular events. Study participants were enrolled in the WHS (Women's Health Study), a recently completed randomized, double-blinded, placebo-controlled clinical trial of low-dose aspirin and vitamin E, in the primary prevention of CVD and cancer in U.S. female health care professionals (33–35). Eligible participants were apparently healthy women, age 45 years or older, who were free of self-reported CVD or cancer at study entry (1992 to 1998) with follow-up for incident CVD through February 2005. At the time of enrollment between 1992 and 1995, participants provided information on whether they were currently taking HT and provided blood samples from which Lp(a) levels were assayed. They also provided demographic, medical history, medication, and lifestyle data, as well as consent for blood-based analyses related to the risk of incident chronic diseases. In total, 27,736 women who provided blood samples for successful Lp(a) analysis, and who also responded to questions about HT, constituted the study population for this analysis.
The study was approved by the institutional review board of the Brigham and Women's Hospital (Boston, Massachusetts).
Baseline plasma measurements
Among WHS participants, 27,736 provided data about HT at baseline and also provided baseline blood samples that were stored in liquid nitrogen (−150°C to −180°C) until the time of analysis. These samples underwent Lp(a) and lipid analysis in a core laboratory certified by the National Heart, Lung, and Blood Institute/Centers for Disease Control and Prevention Lipid Standardization Program. Because of poor agreement of Lp(a) values obtained by different methods (5), we used the only commercially available assay shown by the National Heart, Lung, and Blood Institute and International Federation of Clinical Chemistry Lp(a) standardization groups to not be affected by Kringle IV-Type 2 repeats (5). We determined the concentration of Lp(a) using a turbidimetric assay on the Hitachi 917 analyzer (Roche Diagnostics, Indianapolis, Indiana), using reagents and calibrators from Denka Seiken (Niigata, Japan). The interassay coefficient of variation of Lp(a) concentrations of 17.6 and 58.1 mg/dl were 3.6% and 1.5%, respectively.
High-sensitivity C-reactive protein (CRP) was measured using a validated immunoturbidometric method (Denka Seiken, Tokyo, Japan) (36). Total cholesterol was measured enzymatically. High-density lipoprotein (Roche Diagnostics, Basel, Switzerland) and LDL-C (Genzyme Corporation, Cambridge, Massachusetts) were measured by a homogenous direct method. All lipid determinations were performed on the Hitachi 917 analyzer; 98% of the samples received underwent successful evaluation for each biomarker.
Ascertainment of incident cardiovascular events
All participants were followed prospectively for the occurrence of the composite end point of first ever major cardiovascular event (nonfatal myocardial infarction, nonfatal ischemic stroke, coronary revascularization, or cardiovascular death). Medical records were obtained and reviewed for confirmation of events by 2 cardiologists. Deaths of cardiovascular causes were confirmed by autopsy reports, death certificates, medical records, and contact with family members.
Baseline characteristics such as age, body mass index, smoking status, presence of diabetes, hypertension, family history of premature myocardial infarction, lipid profiles, CRP levels, and post-menopausal status in women on HT, and measures of socioeconomic status such as income and education, were compared with those of women free of HT.
Next, Lp(a) was divided into quintiles on the basis of their distribution in the entire WHS cohort, and these quintile cut points were then used to assess for evidence of interaction between overall Lp(a) quintiles and HT status in Cox proportional hazard models that regressed CVD on Lp(a) quintiles, HT status, and an interaction term of Lp(a) quintiles multiplied by HT status.
Because of evidence of interaction from this evaluation, we recalculated Lp(a) quintile cutoffs separately within each HT strata and then applied these cutoffs to subsequent analysis separately within HT strata. Hazard ratios (HRs), comparing Lp(a) quintiles 2 through 5 with the lowest (referent) quintile, were calculated by Cox proportional hazard models, in models that adjusted first for age (years), then for Framingham covariates as suggested by the most recent National Cholesterol Education Program Adult Treatment Panel III guidelines (37,38), and finally in fully adjusted models that adjusted for age, blood pressure (as defined by Framingham risk models [<120/<75 mm Hg, 120 to 129/75 to 84 mm Hg, 130 to 139/85 to 89 mm Hg, 140 to 159/90 to 94 mm Hg, and ≥160/≥95 mm Hg], diabetes, current smoking status, body mass index (linear continuous), total cholesterol, high-density lipoprotein, CRP, and treatment arms. These variables were chosen a priori for their impact on CVD. Tests for trend across quintiles of Lp(a) were addressed by entering a single ordinal term for each quintile based on the median value for Lp(a) within each quintile.
After stratification by HT, Kaplan-Meier curves were constructed to illustrate cumulative event rates of CVD as a function of Lp(a) quintiles over an average of 10 years of follow up.
Because prior evidence has pointed to increased associations of Lp(a) with CVD in women with concomitant elevations of LDL-C, we assessed for joint associations of Lp(a), LDL-C, and HT with CVD by calculating HRs for CVD by increasing quintiles of Lp(a) in: 1) women with LDL-C ≥ median (121.4 mg/dl) who were taking HT; 2) women with LDL-C ≥ median not taking HT; 3) women with LDL-C < median taking HT; and 4) women with LDL-C < median not taking HT.
To test the strength of our findings, we performed sensitivity analysis using alternative models. Alternative models included those that replaced total cholesterol with LDL, that replaced body mass index with waist circumference, models analyzed within white subjects only, and then models that further adjusted for income (categories included ≤$19,999 per year, $20,000 to $29,900 per year, $30,000 to $39,900 per year, $40,000 to $49,900 per year, $50,000 to $99,900 per year, and ≥$100,000 per year) and education (categories included licensed practical or vocational nurse [LPN, LVN], registered nurse with 2 or 3 years of training [RN 2 year, 3 year], bachelor degree [BS], master, doctorate, or medical degree [MD]). Finally, we divided women taking HT into those taking estrogen only and those taking estrogen plus progesterone combinations, and performed the Cox proportional hazard models for CVD to assess whether the type of HT affected the relationship between Lp(a) and CVD.
All p values are 2-tailed, with a value of <0.05 considered significant. Data analysis was conducted using SAS statistical software version 9.1 (SAS Institute Inc., Cary, North Carolina) and SPLUS version 6.0 (Insightful Corp., Seattle, Washington).
Women taking HT at study initiation were older, had higher total cholesterol, high-density lipoprotein, and CRP levels, and were more likely to be hypertensive compared with women not on HT. They were also more likely to be in higher income brackets and to report longer years of education, but no relationships were seen between income or education and Lp(a) levels. Women on HT also had lower mean low-density lipoprotein (LDL) levels, and were less likely to be diabetic or to be current smokers (Table 1).
Overall assessment of the relationship between Lp(a), HT, and CVD revealed significant interaction between Lp(a) quintiles and HT status on CVD (p interaction = 0.0009).
The different relationships of Lp(a) with CVD, therefore, stratifying for HT therapy status, are shown in Tables 2 to 5.⇓⇓⇓ In women free of HT (Table 2), women in the highest quintile of Lp(a) (≥45.4 mg/dl in HT− strata) were 1.77 times more likely to develop cardiovascular events than those in the lowest quintile (≤3.9 mg/dl; p trend <0.0001) in analyses adjusting for age, smoking, blood pressure, body mass index, cholesterol, high-density lipoprotein, diabetes, hormone use, CRP, and randomization treatment arms. A threshold effect is seen such that risk increased predominantly among women with Lp(a) levels in the highest quintile (66.35 mg/dl) and particularly in women with high LDL (Table 3), as noted previously (7).
In contrast, in women taking HT (Table 4), the relationship of Lp(a) levels to CVD was attenuated, with loss of significance of the relationship between the highest quintile of Lp(a) (≥42.0 mg/dl in women taking HT, with HR of 1.13, top vs. bottom quintile) and CVD (p trend = 0.18 in fully adjusted model), and even modified among women with elevated LDL, in whom synergistic elevations of CVD risk are seen (Table 5).
In sensitivity analyses, further adjustment for income and education, replacement of total cholesterol with LDL, and then replacement of body mass index with waist circumference did not change the effect modification shown graphically in Figure 1, in which high levels of Lp(a) predict CVD in women free of HT, but not in women on HT. The same relationships are seen in analyses limited to Caucasian subjects only.
To address the potential for misclassification of exposure during follow-up, we repeated these analyses censoring follow-up time whenever a change in participant HT status occurred. In this censored analysis, the adjusted HR comparing the highest to lowest quintiles of Lp(a) on risk of all CVD was 1.96 (95% confidence interval [CI]: 1.40 to 2.76), a value if anything larger than that of the uncensored analysis (HR: 1.77; 95% CI: 1.36 to 2.30). In contrast, among women taking HT, the relationship of Lp(a) levels to CVD was attenuated, concordant with the uncensored analysis.
Last, assessing the relationship between Lp(a) and CVD in women taking estrogen versus estrogen plus progestin formulas did not show any differences from the data for HT as a whole (Online Appendixes 1 and 2).
In this prospective study of 27,736 initially healthy women, we found that Lp(a) is a determinant of risk of CVD among women free of HT, but that little such relationship is seen among HT users. This effect modification persists after adjustment for traditional cardiovascular risk factors and socioeconomic variables such as income and education. Quantitatively, the HRs are 1.77 in HT-negative women and only 1.13 in HT-positive women when comparing the top with the bottom Lp(a) quintiles in respective strata. This effect modification is seen even among women with concomitant elevations of Lp(a) and LDL-C, who have the highest incidence of CVD.
We believe that there are at least 2 different explanations for the effect modification seen. For one, it could be attributable to biological interaction between plasma Lp(a) levels and concurrent HT use, which is why it was important to measure HT status at the same time blood was drawn. Alternatively, HT use may be a surrogate variable for healthy lifestyles, not fully captured by the covariates addressed here that ameliorate deleterious relationships of lipid biomarkers with CVD. As for potential direct biological effects, it is intriguing that recent data show that: 1) estrogen may induce increased uptake of Lp(a) by the LDL receptor (39,40); 2) it may cause reduction of Lp(a) production by the liver (30), where Lp(a) production is likely modulated (41–43); and 3) certain cholesterol metabolites may interact directly with HT in the vasculature (44).
Strengths of our study include the reliable measurement of Lp(a), using a reproducible apo(a) isoform-independent assay (5,6) that has complicated many prior studies of Lp(a), the large number of women studied, and the number of cardiovascular end points. Although we consider independence of the assay from apo(a) isoforms a particular strength, it is possible that additional information can be gained by measuring apo(a) isoforms reproducibly. The size of this cohort of women helps us to evaluate what a number of prior epidemiological studies (26–31,45) and randomized controlled trials (46,47), although reporting the association of HT with lower Lp(a) levels, were not powered to assess in relationship to actual hard cardiovascular outcomes. Our data address this question in a primary prevention cohort of women, and confirm trends reported previously in secondary prevention cohorts (32).
Limitations include the predominantly Caucasian nature of our cohort, which does not allow us to assess potentially different relationships of plasma Lp(a) levels and CVD in different ethnic groups (48–50). Also, HT use in women is controversial, and since the publication of the Women's Health Initiative (20) and HERS I (Heart and Estrogen/progestin Replacement Study I) (21) and HERS II (22) studies, the American Heart Association (23), the U.S. Food and Drug Administration (24), and the U.S. Preventive Services Task Force (25) recommend against the use of estrogen and progestin or progesterone for prevention of chronic conditions or for the primary or secondary prevention of CVD (51). However, treatment of menopausal symptoms, although controversial, remains an indication for personalized estrogen use, and new practice standards are emerging to help tailor HRT for menopausal symptoms for short-term use (24,52).
In summary, these data suggest that the predictive utility of Lp(a) is attenuated among women taking HT. The data may inform clinicians' interpretations of Lp(a) values in their patients who are concurrently taking HT.
The authors thank the women who participated in the Women's Health Study and thank Drs. Robert J. Glynn, Kathryn Rexrode, and Emily G. Kurtz for their contributions to the analysis in this study.
For supplementary Tables 1 and 2, please see the online version of this article.
Supported by grants from the Donald W. Reynolds Foundation (Las Vegas, Nevada) and the Leducq Foundation (Paris, France). The Women's Health Study is supported by grants from the National Heart, Lung, and Blood Institute (HL-043851 and HL-80467) and the National Cancer Institute (CA-047988). Dr. Danik had full access to the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Dr. Danik receives support from the National Heart, Lung, and Blood Institute (HL-076443) and the Michael Lerner Foundation. Dr. Rifai receives grant support from Merck Research Laboratories and has served as a consultant to Sanofi-Aventis. He has also received honorarium from Ortho Diagnostics. Dr. Buring has received investigator-initiated research funding and support as a principal investigator from the National Institutes of Health (the National Heart, Lung, and Blood Institute; the National Cancer Institute; and the National Institute of Aging) and Dow Corning Corporation; has received research support for pills and/or packaging from Bayer Heath Care and the Natural Source Vitamin E Association; has received honoraria from Bayer for speaking engagements; and serves on an external scientific advisory committee for a study by Procter & Gamble. Dr. Ridker currently or in the past 5 years has received research funding support from not-for-profit entities including the National Heart, Lung, and Blood Institute; the National Cancer Institute; the American Heart Association; the Doris Duke Charitable Foundation; the Leducq Foundation; the Donald W. Reynolds Foundation; and the James and Polly Annenberg La Vea Charitable Trusts. Dr. Ridker also currently or in the past 5 years has received investigator-initiated research support from for-profit entities including AstraZeneca, Bayer, Bristol-Myers Squibb, Dade-Behring, Novartis, Pharmacia, Roche, Sanofi-Aventis, and Variagenics. Dr. Ridker is listed as a coinventor on patents held by the Brigham and Women's Hospital that relate to the use of inflammatory biomarkers in cardiovascular disease and has served as a consultant to Schering-Plough, Sanofi-Aventis, AstraZeneca, Isis Pharmaceutical, Dade-Behring, and Interleukin Genetics. The funding agencies played no role in the design, conduct, data management, analysis, or manuscript preparation related to this project. The investigators had full access to the data and take full responsibility for its integrity. All investigators have read and agree to the article as written.
- Abbreviations and Acronyms
- C-reactive protein
- cardiovascular disease
- hazard ratio
- hormone replacement therapy
- low-density lipoprotein
- low-density lipoprotein cholesterol
- Received December 28, 2007.
- Revision received March 2, 2008.
- Accepted April 2, 2008.
- American College of Cardiology Foundation
- Lackner C.,
- Cohen J.C.,
- Hobbs H.H.
- Loscalzo J.,
- Weinfeld M.,
- Fless G.M.,
- Scanu A.M.
- Marcovina S.M.,
- Albers J.J.,
- Scanu A.M.,
- et al.
- Marcovina S.M.,
- Koschinsky M.L.,
- Albers J.J.,
- Skarlatos S.
- Shai I.,
- Rimm E.B.,
- Hankinson S.E.,
- et al.
- Bostom A.G.,
- Gagnon D.R.,
- Cupples L.A.,
- et al.
- Rosengren A.,
- Wilhelmsen L.,
- Eriksson E.,
- Risberg B.,
- Wedel H.
- Berglund L.,
- Ramakrishnan R.
- Mosca L.,
- Appel L.J.,
- Benjamin E.J.,
- et al.
- Taskinen M.R.,
- Puolakka J.,
- Pyorala T.,
- et al.
- Roberts W.L.,
- Moulton L.,
- Law T.C.,
- et al.
- Grundy S.M.,
- Cleeman J.I.,
- Merz C.N.,
- et al.
- Tuck C.H.,
- Holleran S.,
- Berglund L.
- Brown S.A.,
- Hutchinson R.,
- Morrisett J.,
- et al.
- Shewmon D.A.,
- Stock J.L.,
- Rosen C.J.,
- et al.
- Christodoulakos G.E.,
- Lambrinoudaki I.V.,
- Panoulis C.P.,
- Papadias C.A.,
- Kouskouni E.E.,
- Creatsas G.C.
- Mosca L.,
- Banka C.L.,
- Benjamin E.J.,
- et al.