| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Department of Psychiatry (K.A.M., J.B., J.C), the Department of Epidemiology (K.A.M., J.B., L.L.S.), and the Department of Psychology (K.A.M.), University of Pittsburgh, Pittsburgh, Pennsylvania; the Department of Preventive Medicine and Behavioral Sciences (S.A.E.-R.), Rush University Medical Center, Chicago, Illinois; and the Department of Epidemiology (M.F.S.), University of Michigan, Ann Arbor, Michigan.
Address correspondence and reprint requests to Karen A. Matthews, Department of Psychiatry, University of Pittsburgh, 3811 O'Hara Street, Pittsburgh, PA 15213. E-mail: matthewska{at}upmc.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Methods: A total of 3292 women enrolled in the Study of Women's Health Across the Nation (SWAN) were followed for five years and had measures of Center for Epidemiologic Studies-Depression and high sensitivity C-reactive protein, Factor VIIc, fibrinogen, plasminogen activator inhibitor Type 1(PAI-1), and tissue-type plasminogen activator antigen (tPA-ag) up to four times during the follow-up period. Women were pre- or early perimenopausal status at study entry and were of Caucasian, African American, Hispanic, Japanese, or Chinese race/ethnicity.
Results: Unadjusted longitudinal mixed regression models showed that over a 5-year period, higher depressive symptoms were related to higher fibrinogen, PAI-1, and tPA-ag levels, all p < .0001. Taking into account health history, medication use, ethnicity, aging, and menopausal status, the depressive symptoms were related to fibrinogen, p < .01, and PAI-1, p < .05. Depressive symptoms were related only to fibrinogen in models that also included body mass index, p < .05.
Conclusions: Depressive symptoms may be associated with cardiovascular risk in perimenopausal women in part through hypercoagulability. This is the first study to test the association of depressive symptoms and hemostatic and inflammatory markers across time.
Key Words: depression • inflammation • hemostatis • menopause • women • longitudinal
Abbreviations: CHD = coronary heart disease; PAI-1 = plasminogen activator inhibitor Type 1; hs-CRP = high sensitivity C-reactive protein; BMI = body mass index; SWAN = Study of Women's Health Across the Nation; CES-D = Center for Epidemiologic Studies-Depression; tPA-ag = tissue-type plasminogen activator antigen; IL = interleukin.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Among the potential pathways that may be relevant to understanding depression's risk for CHD morbidity and mortality is chronic inflammation (4), which is thought to be important throughout the natural history of coronary atherosclerosis from the initial defect in endothelial function through plaque rupture (5). Markers of chronic inflammation are C-reactive protein (CRP) and fibrinogen. An imbalance in the coagulation and fibrinolytic systems leading to platelet activation and aggregation is another potential pathway linking depression and CHD. Markers of an imbalance on the coagulation side are increased levels of Factor VII and fibrinogen. Fibrinogen is also the major determinant of plasma and blood viscosity. On the fibrinolytic side are the markers of tissue plasminogen activity (tPA), the main fibrinolytic stimulator, and plasminogen activator inhibitor Type 1 (PAI-1), the primary inhibitor of the fibrinolytic process. Impaired fibrinolytic function may be reflected in high plasma levels of PAI-1 or tPA or low levels of activation products like D-dimers (6). The biological plausibility of such pathways connecting depression and increased CHD risk rests on evidence showing that stress systems are associated with increased coagulation and alterations in fibrinolysis, although inflammatory responsivity may also contribute to the occurrence of depressive symptoms (7,8).
A number of epidemiological studies have reported the concurrent associations between depression and hemostatic and/or inflammatory markers. In the Cardiovascular Health Study of the elderly, depressive symptoms were associated with levels of CRP, Factor VII, fibrinogen, white blood cells, and platelet count (9). However, adjustments for a number of covariates, including body mass index (BMI), reduced the associations to nonsignificance. Similarly, in the Rotterdam Study of the elderly (10), depressive disorders were not related to CRP. However, in the Health ABC Study, depressive symptoms were associated with CRP in fully adjusted models, with the association of depressive symptoms and interleukin (IL)-6 stronger in men than in women (11). In younger samples, depressive and anxious symptoms were not associated with CRP, fibrinogen, IL-6, tumor necrosis factor-
, and lymphocytes in healthy participants from Whitehall II (12), although there was an association with endothelial dysfunction. Life-time recurrent depression was associated with CRP in men, but not in women in the Third Annual National Health and Nutrition Examination Survey (13,14). However, among both men and women in the ATTICA study (15), depressive symptoms were related to CRP, white blood cell count, and fibrinogen levels in multivariate models. Some evidence suggests that depression is related concurrently to hemostatic and inflammatory markers, with the evidence weaker in women than in men. There are no longitudinal data yet available.
Obesity and central adiposity may affect the association between depression and some inflammatory markers. Miller et al. (16) reported in a small sample that clinical depression is linked cross-sectionally with inflammation through central adiposity and leptin levels. In men enrolled in the study monitoring trends and determinants of cardiovascular diseases (17), depression and anxiety were related to CRP but only among those who were obese. Among Army personnel, depression was associated with CRP, with adjustments for body mass index (BMI) reducing the associations to nonsignificance (18). Taken together, available studies suggest that the links between inflammatory factors and depression may be influenced by obesity.
The present paper tests the hypothesis that women's depressive symptoms are related to five markers of hemostasis and inflammation: a) Factor VIIc, b) tissue-type plasminogen activator antigen (tPA-ag), c) PAI-1, d) fibrinogen, and e) high sensitivity C-reactive protein (hs-CRP). We report longitudinal data of >3000 women followed for 5 years who were enrolled in the Study of Women's Health Across the Nation (SWAN). SWAN enrolled Caucasian, African American, Japanese, Chinese, and Hispanic premenopausal and early perimenopausal women. Secondarily, we explore a) the role of obesity in understanding any observed relationships and b) the influence of menopausal status.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Of the 3302 women, five were excluded because of pregnancy during the follow-up period and five because either the Center for Epidemiologic Studies Depression (CES-D;20) (n = 2) or hemostatic/inflammatory markers (n = 3) were missing at all examinations, resulting in an analytic sample of 3292 women. Longitudinal analyses were based on a maximum total of 3249 women because of missing BMI data at all examinations (n = 12) or missing education data (n = 31). Analytic sample size varied also because of missing data at all examinations (missing were 8 for hs-CRP, 11 for Factor VIIc, 12 for fibrinogen, 25 for PAI-1, and 26 for tPA-ag). The study retention rate at the end of the fifth follow-up examination was 79%.
Procedures
SWAN participants at all seven sites were assessed at study entry (baseline) and once a year with a common protocol. All study forms and materials were available in English, Spanish, Japanese, and Cantonese and bilingual staff was used, as appropriate. Baseline and annual assessments included interviewer administered and self-administered questionnaires about health, lifestyle, and psychosocial factors. Anthropometric measurements and phlebotomy procedures were obtained with standardized protocols. At each examination, the fasting blood draw was targeted to the follicular phase of the menstrual cycle (days 2–5) in menstruating women and before 10 AM. All samples were maintained at 4°C until separated and then were frozen at –80°C and shipped on dry ice to a central laboratory. For budgetary reasons, assays were completed for hemostatic/inflammatory factors only at baseline, and years 1, 3, and 5. Thus, this report is based on hemostatic and inflammatory measures for a maximum number of four examinations per woman across a 5-year period.
Measures
Hemostatic Factors/Inflammatory Markers
Outcome variables included hs-CRP, Factor VII activity, fibrinogen, PAI-1 and tPA-ag measured in plasma. The hs-CRP was quantitated using an ultrasensitive rate immunonephelemetric method (hs-CRP, Dade-Behring, Marburg, Germany). Fibrinogen and Factor VIIc activity were measured in frozen citrated plasma (MLA ELECTRA 1400C, Medical Laboratory Automation Inc., Mt. Vernon, NY) using a turbidometric detection system. PAI-1 was measured with a sandwich procedure using a solid phased monoclonal antibody and an enzyme labeled goat second antiserum for detection (IMUBIND plasma PAI-1 ELISA, American Diagnostica, Greenwich, CT). The tPA-ag was measured in plasma using a double antibody in an enzyme linked immunosorbant assay (IMUBIND tPA ELISA, American Diagnostica, Greenwich, CT). The assay uses human single chain t(PA) as a standard calibrated against an international standard (NIBSAC, Hertfordshire, UK). For analyses, blood samples were recorded as occurring between 8:00 AM and 10:00 AM (morning blood draw) versus not; and whether or not the participant had fasted 12 hours before the blood draw.
Depression
Depressive symptoms were assessed at baseline and annually with the CES-D Scale, a 20-item measure that asks about the frequency of being bothered by depressive symptoms during the previous week on a scale of 0 (rarely) to 3 (most or all of the time) for a total range of 0 to 60 (20). This scale was developed to screen for clinical depression in community samples. The measure has been used in multiethnic cohorts, including African, Chinese, Japanese, and Hispanic American women (21–23). The CES-D and the other instruments in SWAN were translated into Cantonese, Japanese, and Spanish and were offered in either English or the participant's native language. In our sample, Cronbach's coefficient was 0.90.
Other Variables
Age, race/ethnicity, education (
high school degree, some college/vocational training, college degree, or more), and total physical activity during housework and leisure activities (24) were obtained at the baseline examination. Medication use and health history were obtained at each examination. To coincide with timing of the assays, responses for follow-up examinations 2 and 3 and for examinations 4 and 5 were combined for analyses of health conditions; e.g., stroke reported at either examination 2 or 3 was considered as a stroke for the analysis of year 3 data. Questions on medication use after the last study examination included checking use of steroids, insulin, and antihypertensive medication, anticoagulants, and heart medications (e.g., warfarin, digoxin, nitroglycerin). We defined stroke/heart condition as self-reported stroke, heart attack, angina, or taking anticoagulants or heart medication since the previous examination.
BMI, smoking, exogenous hormone use, menopausal status, and medication for nerves/depression were assessed annually. BMI (kg/m2) was calculated from measurements of weight and height, which were obtained with a calibrated scale and a standiometer. Smoking status was assessed as current versus not. Menopausal status was based on menstrual bleeding patterns in the previous 12 months and was categorized as: a) premenopausal = menstrual period in the past 3 months with no change in regularity in the past 12 months; b) early perimenopausal = menstrual period in the past 3 months with some change in regularity over the previous 12 months; c) late perimenopausal = no menstrual period within the past 3 months, but some menstrual bleeding within the past 12 months; d) post menopausal = no menstrual period within the past 12 months; e) surgical menopause = hysterectomy or bilateral oophorectomy; f) indeterminant menopausal status = used hormone therapy before final menses or surgical menopause; thus, status could not be determined. Based on SWAN eligibility requirements, all women were pre- or early perimenopausal at baseline.
Data Analyses
Except for fibrinogen, which was normally distributed, the outcome variables were log transformed because of their skewed distributions. Because some women had levels of hs-CRP and PAI-1 between zero and one, a factor of one was added to these markers before transformation to avoid negative log values. The hs-CRP values >10 were excluded from all analyses (7% of all hs-CRP values). The CES-D scores were also not normally distributed and were log transformed after adding one. Longitudinal associations between increased CES-D scores and inflammatory/hemostatic markers across the four examinations were evaluated using longitudinal linear mixed regression models. These models included a (woman-specific) random intercept, which enabled us to ascribe a "women-specific" interpretation to model parameters (as opposed to the "population-averaged" interpretation). Exploratory analyses and standard model fit statistics indicated that an auto-regressive error correlation structure was more suitable than simpler alternatives (25). The auto-regressive error structure allows observations that occur closer in time to be more strongly correlated. Mixed regression models do not require complete data on all participants across all examinations and are robust to most types of missing data. Diagnostic plots were used to verify model assumptions of normality. Scatter plots and ordinal analyses showed no evidence of nonlinear trends; thus, linear analyses were used.
Study site, race/ethnicity, baseline age, and time after baseline (aging) were included in each model. Covariates for the multivariable models were determined a prior based on the literature and included education and time-covarying covariates of cardiovascular condition (self-reported stroke, heart attack, angina, anticoagulants, and heart medications), other medications (insulin, steroids, antihypertensive, "nerves/depression" pills, hormone therapy), smoking status, blood draw occurring in the morning window, and menopausal status. Because of our interest in the role of BMI, multivariable models were calculated with and without BMI in the models as well as tests for interaction of CES-D scores and BMI. Continuous (log transformed) CES-D scores were included as the independent variable in all models. To coincide with the collection of hemostatic and inflammatory markers, data from examinations 0, 1, 3, and 5 were used for the covariates and CES-D. We also conducted a more traditional analysis where we examined whether baseline CES-D scores predicted the hemostatic/inflammatory markers over the 5-year period, including physical activity and other covariates measured at baseline only. Analyses were computed using SAS (Version 8, SAS Institute, Inc., Cary, NC).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
|
|
Associations Between Depressive Symptoms and Hemostatic/Inflammatory Markers Over 5 Years of Follow-Up
Women with higher depressive symptoms had significantly higher levels of fibrinogen, PAI-1, and tPA-ag across all examinations in unadjusted longitudinal linear mixed models, p < .00001 (Table 3, row 1). Fully adjusted models showed that high depressive symptoms were positively and significantly associated with fibrinogen across examinations (p = .05; Table 3, row 2). When all covariates remained in the model except for BMI (second to last row Table 3), the associations between higher CES-D scores and higher fibrinogen and PAI-1 were significant. Tests for a moderating effect of BMI showed a significant interaction for only PAI-1, p = .01, such that higher CES-D scores were more strongly associated with higher PAI-1 among women with a lower BMI. High CES-D scores at baseline (adjusted for other baseline covariates including physical activity) were unrelated to any of the outcomes across the follow-up period (data not shown).
|
Associations Between Relevant Covariates and Inflammatory/Hemostatic Markers
Several associations between covariates and markers are noteworthy. The strong positive associations between BMI and the hemostatic/inflammatory markers were observed throughout the follow-up period (Table 4). Current smokers had increased markers except for Factor VII-c. Baseline physical activity was associated with all markers, except for a marginal association with Factor VII-c (data not shown).
|
With aging (time variable), hs-CRP and Factor VIIc increased and fibrinogen, PAI-1, and tPA-ag decreased, independent of menopausal status change. Menopausal status was an important predictor of all markers, except hs-CRP, independent of aging. Compared with premenopausal women, late perimenopausal women had increased Factor VIIc, fibrinogen, and tPA-ag. PAI-1 and tPA-ag were increased among women of indeterminant status (usually because they were using hormone therapy before the last menses) and tPA-ag and fibrinogen were also increased among postmenopausal women compared with premenopausal women. Concurrent hormone use was associated with higher hs-CRP and Factor VII and lower fibrinogen, PAI-1, and tPA-ag.
Ethnicity was strongly associated with markers. African Americans had higher fibrinogen and lower Factor VII and PAI-1, whereas Chinese and Japanese had higher PAI-1 and tPA-ag, and Hispanics had higher tPA-ag, relative to Caucasians.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
As is reported elsewhere (26–28), we found that BMI is an important correlate of all the hemostatic and inflammatory markers that we measured. Some have suggested that adiposity is an important mediator of the relationship between depression and inflammation, specifically hs-CRP, whereas others suggested that it is a moderator. In our sample, CES-D scores were not associated with BMI. Thus, BMI was not a candidate mediator. We did find that statistical adjustments for BMI reduced the association of PAI-1 and CES-D scores to nonsignificance and that CES-D scores had a stronger association with PAI-1 for those with lower BMI. This suggests that BMI might play a role (along with other factors) in accounting for the associations between depressive symptoms and activated fibrinolysis.
It is noteworthy that depressive symptoms throughout the 5 years were associated with increased fibrinogen in the multivariate models that included BMI. Fibrinogen is a unique marker as it is both an acute-phase reactant, like hs-CRP, and binds with platelet glycoprotein IIb/IIIa to facilitate platelet aggregation. It is an important determinant of blood and plasma viscosity. Our positive fibrinogen findings from the longitudinal analyses differ from the null cross-sectional associations reported in the Cardiovascular Health Study of the elderly (9) and the Whitehall Study of young persons (12), but are consistent with the positive cross-sectional associations reported in the ATTICA Study (15).
Because the women in our study were traversing the menopausal transition, we included menopausal status and use of hormone therapy, independent of age at study entry and chronological aging thereafter, in the models. Late perimenopausal women had elevated Factor VIIC, fibrinogen, and tPA-ag levels, with fibrinogen and tPA-ag levels also being elevated among postmenopausal women, all compared with premenopausal women in fully adjusted analyses. These findings suggest that the late perimenopausal stage may be uniquely associated with alterations in the hemostasis. Confidence in the validity of our findings is enhanced by associations between current hormone use and elevated hs-CRP and lowered fibrinogen levels, observations previously reported in other samples (29). It has been suggested that the elevated hs-CRP levels associated with concurrent hormone use is one factor that may explain the unexpected associations of hormone therapy and increased risk for CHD in clinical trials. The above findings should be considered tentative and must be confirmed when more women enrolled in SWAN have transitioned. This is because the findings are based on women who transitioned early and early transitioners are known to be less educated and more often smokers, both factors that are associated with CHD risk (30,31).
We have debated among ourselves about whether the analytic approach we have used is the best test of the study hypotheses. The inclusion of a large number of covariates as well as time varying covariates that are theoretically and empirically associated with hemostatic and inflammatory markers may have overcontrolled for depressive symptoms. For example, increased depressive symptoms may prevent current smokers from contemplating quitting or may change past smokers into current smokers, masking an effect of symptoms. We adjusted for use of medications for "depression/nerves," which are significantly associated with CES-D scores at all years cross-sectionally. We presented the full multivariate and univariate analyses so that readers can understand the basis of our study conclusions. We also debated whether our study hypotheses are best tested in women with no history of cardiovascular disease, no medications that influence the outcome variables, and nonsmokers. In addition to losing power to test the hypotheses, we also believed that any results from such an analysis would be less generalizable and would not lead us to meet our ultimate goal of drawing conclusions about the likely pathways connecting depressive symptoms and incident and recurrent CHD and stroke. Finally, our debate centered on whether our hypotheses are best tested by considering depressive symptoms continuously or considering the effect above a specific threshold of CES-D, usually considered as indicative of mild depressive symptoms. We opted for the former, given that we are not testing the effect of clinical depression in the analysis and that the hypotheses do not propose a nonlinear effect. Further, as noted earlier, examination of scatterplots and other tests for nonlinearity did not show nonlinear effects.
Our study has a number of strengths and limitations. Regarding limitations, our study did not systematically code concurrent data at each examination on nonprescription drug use or acute illness that might affect the markers. Second, only 79% of women were retained in the analysis by the fifth annual examination and we know that women who had only baseline data reported more depressive symptoms and higher inflammatory/hemostatic markers than women who remained in the study for at least two examinations (data not shown). The analytic approach that we used, however, adjusted for missing data in estimating the effects of depressive symptoms on markers. The strengths of the study include the longitudinal design, the large and ethnically diverse sample of women, the breadth of the measures of hemostatic and inflammatory markers, the repeated assessments of both the predictor and outcome variables, and the well characterized covariates known to influence hemostatic and inflammatory markers.
In summary, among women approaching menopause, depressive symptoms are related over a 5-year period to elevated fibrinogen, PAI-1, and tPA-ag levels. Taking into account health history, medication use, ethnicity, aging, and menopausal status, depressive symptoms are related to fibrinogen and PAI-1. Weight adjusted for height is a strong determinant of all the markers, and depressive symptoms are related more strongly to PAI-1 among women with lower BMI. Depressive symptoms may be associated with CHD risk, in part, through hypercoagulability and less likely through inflammation; however, the magnitude of the effect is not large. Our conclusions do not apply to the CHD risk associated with clinical depression and cannot be generalized to men or late postmenopausal women. Our study is the first to report the association of depressive symptoms and hemostatic and inflammatory markers across time.
Clinical Centers: University of Michigan, Ann Arbor, MI: MaryFran Sowers, PI; Massachusetts General Hospital, Boston, MA: Robert Neer, PI, 1994–1999; Joel Finkelstein, PI, 1999–Present; Rush University, Rush University Medical Center, Chicago, IL: Lynda Powell, PI; University of California, Davis/Kaiser, Oakland, CA: Ellen Gold, PI; University of California, Los Angeles, CA: Gail Greendale, PI; University of Medicine and Dentistry-New Jersey Medical School, Newark, NJ: Gerson Weiss, PI, 1994 to 2004; Nanette Santoro, PI, 2004–Present; and the University of Pittsburgh, Pittsburgh, PA: Karen Matthews, PI.
NIH Program Office: National Institute on Aging, Bethesda, MD: Marcia Ory, 1994–2001; Sherry Sherman, 1994–Present; National Institute of Nursing Research, Bethesda, MD: Program Officers.
Central Laboratory: University of Michigan, Ann Arbor, MI: Daniel McConnell; Central Ligand Assay Satellite Services.
Coordinating Center: New England Research Institutes, Watertown, MA: Sonja McKinlay, PI, 1995–2001; University of Pittsburgh, Pittsburgh, PA: Kim Sutton-Tyrrell, PI, 2001–Present.
Steering Committee: Chris Gallagher, Chair; Susan Johnson, Chair.
We thank the study staff at each site and all the women who participated in SWAN.
|
NOTES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
The Study of Women's Health Across the Nation (SWAN) has grant support from the National Institutes of Health (NIH), Department of Health and Human Services (DHHS), through the National Institute on Aging, the National Institute of Nursing Research and the NIH Office of Research on Women's Health (Grants NR004061; AG012505, AG012535, AG012531, AG012539, AG012546, AG012553, AG012554, AG012495).
DOI:10.1097/01.psy.0000256574.30389.1b
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  S. A. Everson-Rose, T. T. Lewis, K. Karavolos, S. A. Dugan, D. Wesley, and L. H. Powell Depressive Symptoms and Increased Visceral Fat in Middle-Aged Women Psychosom Med, May 1, 2009; 71(4): 410 - 416. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. G. Lee, M. C. Carr, S. J. Murdoch, E. Mitchell, N. F. Woods, M. H. Wener, W. L. Chandler, E. J. Boyko, and J. D. Brunzell Adipokines, Inflammation, and Visceral Adiposity across the Menopausal Transition: A Prospective Study J. Clin. Endocrinol. Metab., April 1, 2009; 94(4): 1104 - 1110. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Hamer, G. J. Molloy, and E. Stamatakis Psychological distress as a risk factor for cardiovascular events: pathophysiological and behavioral mechanisms. J. Am. Coll. Cardiol., December 16, 2008; 52(25): 2156 - 2162. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. C. Stewart, D. Janicki-Deverts, M. F. Muldoon, and T. W. Kamarck Depressive Symptoms Moderate the Influence of Hostility on Serum Interleukin-6 and C-Reactive Protein Psychosom Med, February 1, 2008; 70(2): 197 - 204. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
