| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Department of Behavioral Sciences and Community Health (J.A.W., G.A.), Department of Community Medicine (H.T.), and Department of Medicine, Section of Hypertension (G.A.M.), the University of Connecticut Health Center, Farmington, CT.
Address correspondence and reprint requests to Julie Wagner, PhD, Department of Behavioral Sciences and Community Health MC3910, University of Connecticut Health Center, 263 Farmington Ave., Farmington, CT 06410. E-mail: juwagner{at}uchc.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSNOTESREFERENCES |
|---|
Methods: Thirty-nine postmenopausal women with no known or suspected cardiovascular disease participated. Nineteen had a positive lifetime history of major depressive disorder, and 20 were never depressed. None were currently depressed, and all had been free of major depressive disorder and antidepressant medications for 
1 year. History of depression was assessed with the Structured Clinical Interview for DSM-IV, enhanced by a modified version of the timeline follow-back method. Current depressive symptoms were measured with the Center for Epidemiological Studies Depression scale (CES-D). Brachial artery FMD was measured with ultrasound and calculated as percent dilation from baseline.
Results: After controlling for current subclinical depressive symptoms, ethnicity, hormone replacement therapy, and presence of the metabolic syndrome, previously depressed women showed significantly and clinically meaningful lower FMD than never depressed women. Controlling the same covariates, there was a dose-response relationship between number of depressive episodes and FMD. Examination of FMD means showed a significant negative correlation between number of depressive episodes and FMD.
Conclusion: In postmenopausal women, depression continues to show a negative relationship to endothelial functioning even after years of remission. This relationship is not accounted for by residual depressive symptoms. Implications pertain to exclusion of previously depressed persons from control groups in research exploring the relationship between depression and cardiovascular disease.
Key Words: depression • measurement • endothelial function • coronary heart disease • women • postmenopausal
Abbreviations: CHD = coronary heart disease; FMD = flow-mediated dilation; CES-D = Center for Epidemiological Studies Depression questionnaire; BMI = body mass index; SCID = Structured Clinical Interview for DSM-IV; TLFB = timeline follow-back; HRT = hormone replacement therapy.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSNOTESREFERENCES |
|---|
Major depressive disorder is a risk factor for incident CHD (4), and women have higher rates of major depressive disorder than their male counterparts (5,6). Although less is known about depression as a risk factor for CHD in women and minorities relative to men and whites, the Women’s Health Initiative did show that depression is a risk factor for incident CHD in women specifically (7). It has been hypothesized that depression may exert its deleterious effects at least in part via endothelial dysfunction (8). Currently depressed persons show altered chemical markers of endothelial function and immune activation, inflammatory response, and coagulation (9–17). Functional alteration in the endothelium relates to impaired hyperemic response to experimental arterial occlusion (flow-mediated dilation, or FMD). Presently there are three controlled studies of current major depressive disorder and FMD, each of which showed that current depression or its symptoms were associated with attenuated brachial artery FMD (18–20), although the temporal direction of this association is not clear.
Far less is known about the effect of history of depression. Specifically, it is unknown whether endothelial dysfunction continues after the depressive episode resolves. Data from other areas of research suggest that this might be the case. In the Women’s Health Initiative, survival curves for those depressed and not depressed become more separate over time, indicating that the deleterious association between depression and vascular disease is not proximal to depression assessment (7). Major depressive disorder is associated with cellular alterations that remain abnormal in remission, thus creating risk for worse outcomes long after the depressive episode (21,22). This may help to explain why patients with a history of major depressive disorder before myocardial infarction have higher mortality than those who experience their first episode postinfarct (23). Certainly there are state-dependent depression effects which are perhaps even stronger than any effects that might be detected during depression remission. However, history of depression has been associated with health problems in middle-aged women, including physical symptoms (24), coronary and aortic calcification (25), and carotid plaque (26).
Moreover, we are unaware of any data relating endothelial function to duration of depression over the lifespan. Specifically, it is unknown whether a greater number of depressive episodes are related to more impairment in endothelial function. Major depressive disorder can be a chronic, progressive condition, with each episode increasing the likelihood of an additional episode and subsequent interepisode recovery periods of shorter duration with decreased symptom relief (6,27). It is possible that there is a "dose-response" relationship between depression and cardiovascular risk such that more depressive episodes and/or longer exposure to depression is associated with higher risk.
The purpose of this study was two-fold. Our primary goal was to determine whether the association between depression and endothelial function persists into depression remission among postmenopausal women. In order to do this, we did not study currently depressed women. Instead, we compared never depressed women to currently nondepressed women who had a history of major depressive disorder. Our secondary goal was to explore whether there was a dose-response relationship between number of depressive episodes and endothelial functioning. In addressing both goats, we took into account current depressive symptoms in order to isolate the effects of the depressive episodes.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSNOTESREFERENCES |
|---|
1 year before enrollment. Exclusion criteria included current depressive disorder, known or suspected CHD (angina, angiography, positive stress test, heart failure, or major adverse cardiac event), stroke, peripheral vascular disease, diabetes mellitus, surgical menopause, current psychoactive medication, active substance abuse/dependence, impaired mental status, and current smokers. Current depressive disorder was selected as exclusion criteria in order to ensure that any effects of history of depression could not be better accounted for by current major depressive episode or current depressive symptoms. Current psychoactive drugs, including antidepressants, were selected because their effect on endothelial functioning is largely unknown. Known or suspected vascular disease was selected because we were interested in studying depression as a primary risk factor, i.e., as a risk factor for development of new-onset CHD in women without the disease. Smoking and diabetes were selected because they are strongly related to endothelial function (e.g., 28,29). Surgical menopause was selected because women who undergo double oophorectomy have hormonal alterations, either by virtue of hormone deficiency or replacement, that may affect their cardiovascular health (3). Women with a history of depression were deliberately oversampled. Women were recruited using newspaper advertisements, listserves, and state employee paycheck inserts.
Measures
Demographics were obtained via self-report questionnaire. A research nurse took a medical history which included self-reported age; smoking history; as well as use of lipid-lowering, antihypertensive, antioxidant, aspirin, and hormone replacement therapies (HRT). Women bring all medications with them to the study visit to decrease unreliability due to poor recall. Questionable data were confirmed with the participant’s physician. Caffeine consumption was measured via self-report as servings of coffee, tea, and caffeinated soda per typical day.
Presence of the metabolic syndrome was determined according to NCEP ATPIII criteria for women (30,31). Waist circumference was measured by the study nurse by taking measurements at the midpoint between the upper iliac crest and the lower costal margin in the midaxillary line. Blood pressure was measured per JNC-VII guidelines with manual sphygmomanometer by the research nurse. Body mass index (BMI) was calculated from weight and height data as weight in kilograms divided by height in meters squared. Weight in kilograms and height in centimeters are measured by the research nurse with calibrated equipment. Fasting plasma glucose is analyzed in the University of Connecticut’s hospital clinical lab with the Beckman Coulter LXI system. Physical activity was assessed via the Framingham Physical Activity Index, which provides self-reported physical activity during a typical 24-hour period (32).
Main Study Variables
Brachial artery FMD was assessed with ultrasound. Flow is increased by cuff occlusion of the forearm and then releasing the cuff. The method is repeated using an exogenous nitric oxide donor (nitroglycerine) to assess endothelium-independent activity. Images are captured on high-resolution videotape by a blinded technician and analyzed offline by a blinded observer in four stages. (a) Artery baseline diameter is obtained from five measurements at diastole in each of three frames. (b) Artery diameter during hyperemia is measured similarly on the 1-minute post hyperemia images. Reactive hyperemia is calculated and reported here as the percentage increase in brachial artery diameter from baseline.
Current depressive symptoms were measured with the Center for Epidemiological Studies Depression questionnaire (CES-D; 33). This widely used 20-item scale measures depressive symptoms during the preceding week. The scale has high internal consistency, test-retest reliability, and construct validity in both community and clinical samples (34,35).
History of depression was assessed using the Structured Clinical Interview for DSM-IV (SCID; 36). History of depression was measured as a dichotomous variable (yes versus no), and as a continuous variable (number of depressive episodes over the lifespan). Interrater and test-retest agreement in the "excellent" range (37) for presence or absence of current Axis I disorders is not unusual with the SCID (38). In order to enhance recall, we adapted the timeline follow-back (TLFB) for use in this study (39). TLFB is a measure of previous behavior widely used in behavioral studies. In our version of the TLFB interview, several memory aids were used to enhance recall (e.g., key dates and life events served as anchors for reporting behavior, and a visual timeline calendar was constructed with the participant to graphically represent this information). The TLFB has been used with various populations to study a variety of psychological phenomena (40–46). The TLFB has demonstrated high test-retest reliability, content validity, concurrent criterion validity, collateral (informants) validity, and construct validity in both general and clinical samples (47). Validity of the method is highest when the respondent is willingly engaged in the interview procedure, as were participants in our study (48).
SCID Reliability
Four doctoral-level psychologists assessed lifetime history of major depressive disorder using the SCID. Audiotapes of 12 randomly selected SCID administrations showed 94% interrater reliability for individual SCID items (range, 60% to 100%), 100% interrater reliability for presence/absence of past major depressive disorder, and 100% interrater reliability for number of past depressive episodes. It should be noted that although checking interrater reliability in this fashion accounts for interindividual scoring of data obtained by a single SCID administration, it does not account for potential interindividual differences in SCID administration.
Statistical Analyses
Analyses were conducted using SPSS version 12.0. In order to test the association between history of depression and FMD, group differences were analyzed using ANCOVA, with depression history treated as a dichotomous independent variable (yes versus no) and FMD as a continuous dependent variable. In order to test a dose-response relationship between number of depressive episodes and FMD, Pearson’s correlations and partial correlations were calculated.
Results of evaluations of assumptions revealed that FMD data were normally distributed both within and among depression history groups. According to guidelines of Tabachnick and Fidell (49), the ratio of cases to predictors was satisfactory. Inspection of the data revealed no outliers for predictors or FMD. Cronbach’s 
for the CES-D was satisfactory, = 0.74.
Selection of Covariates
2 showed no differences by depression status for ethnicity, education, income, smoking history, daily vitamins, daily aspirin, HRT, antihypertensives, lipid-lowering agents, or presence of the metabolic syndrome. Point biserial correlations showed no relationship between history of depression and age, BMI, waist circumference, physical activity, systolic BP, diastolic BP, or caffeine consumption. The point biserial correlation between history of depression and CES-D scores was significant, r = 0.48, *p < .01. Based on these results and review of relevant literature, and to control for current levels of distress, CES-D scores were entered as a covariate (20).
In addition to CES-D scores, we also investigated neuroticism as an indicator of current distress using the neuroticism subscale of the NEO (50–52). Zero-order Pearson correlations revealed no relationship between neuroticism and FMD. Point biserial correlation showed a positive association between history of depression and NEO scores, r = 0.34, *p < .05. Inclusion of neuroticism as a covariate did not change the direction or significance of any of our findings, so results are reported without neuroticism as a covariate.
Looking at the sample as a whole, zero-order Pearson’s correlations showed no significant relationship between FMD and age, BMI, caffeine intake, physical activity, waist circumference, or CES-D scores. ANOVA showed no group differences in FMD by ethnicity. Point biserial correlation showed no relationship between FMD and antihypertensive agents, lipid-lowering agents, daily aspirin, daily vitamins, metabolic syndrome, smoking history, or HRT. Based on review of literature with postmenopausal women, ethnicity (53) and HRT (20) were entered as covariates. Based on a broader review of the literature, metabolic syndrome was entered as a covariate because it is a single measure that incorporates adiposity, glucose dysregulation, blood pressure, and lipids, all of which have been linked to endothelial function.
Procedures
The following procedure was approved by the University of Connecticut Health Center IRB. On day 1, a psychologist and study nurse met the participant at the University of Connecticut Health Center general clinical research center. After obtaining informed consent, the nurse completed the medical history and brief physical examination, the psychologist administered the SCID, and the participant completed the Framingham Physical Activity questionnaire and the CES-D. The participant was instructed how to prepare for day 2, which typically occurred 1 week later. For 24 hours before day 2, women refrained from exercise and high-fat foods, lipid-lowering and antihypertensive medications, as well as aspirin, vitamins, and caffeine. For 12 hours before day 2, women fasted. On the morning of day 2, a blood sample was obtained, and the participant completed the brachial artery FMD study, was served breakfast, and was financially compensated $75 for participation.
Arterial ultrasound study of endothelium-dependent FMD was performed according to the 2002 guidelines set forth by the International Brachial Artery Reactivity Task Force (54). The procedure took place in a temperature-controlled room (22°C) with continuous EKG tracing. Patients rested supine for 10 minutes. Blood pressure and heart rates were measured every 3 minutes in the free arm. The dominant arm was used for the study and was immobilized supine. A 12-MHz linear array transducer was used. The studies were divided into parts A and B. Part A: Longitudinal diastolic gated images of the brachial artery were obtained above the elbow, and a straight segment was used for image capture. Transducer beam was placed perpendicular to the posterior wall intimal surface. Baseline images were taken for 20 seconds, as well as diastole-gated images (five cycles). Pulsed flow systolic Doppler velocities were measured at the center of the vessel, with the transducer at 60 to 70 degrees. The exact position of the transducer was measured from the elbow and from the medial and lateral epicondyles for the second study. Part B: A blood pressure cuff was placed distal to the study site and inflated 10 mm Hg above systolic blood pressure for 5 minutes and then released. Doppler systolic blood flow was taken at 1 and 2 minutes, and artery images were repeated at 30 and 60 seconds after deflation. After a 20-minute period of rest, a repeat baseline study was obtained, followed by measurements 3 minutes after sublingual nitroglycerine 0.3 mg.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSNOTESREFERENCES |
|---|
|
In women with a history of depression, the average age of onset of major depressive disorder was 36 years, and 10 years had passed since remission of the most recent episode. The majority had experienced recurrent major depressive episodes (69%). By design, all had been free of major depression and antidepressant medication for at least 12 months. See Table 2 for depression descriptives.
|
History of Depression
A one-way ANCOVA was performed with previous major depressive disorder (yes/no) as the independent variable and FMD as the dependent variable. As outlined previously, covariates were CES-D scores, ethnicity, HRT, and presence of the metabolic syndrome. Previously depressed women demonstrated poorer dilation (M = 4.4%) than never depressed women (M = 11.2%), F (5, 39) = 4.83, *p < .05. None of the covariates were significantly associated with FMD. A one-way ANOVA with nitroglycerine-induced endothelium-independent FMD as the dependent variable showed no significant difference by history of depression.
Dose-Response Relationship Between Exposure to Depression and FMD
Zero-order Pearson’s correlation between FMD and number of depressive episodes was significant, r = –0.34, *p < .05. Examination of mean FMD scores suggests a dose-response relationship between number of depressive episodes and FMD, zero episodes M = 10.3%, SD = 0.02, one episode M = 7.5%, SD = 0.03, two episodes M = 5.0%, SD = 0.03, and three episodes M = 2.6%, SD = 0.04. Adjusting for CES-D scores, ethnicity, HRT, and presence of the metabolic syndrome did not attenuate this relationship, partial r = –0.36, *p < .05. Examination of adjusted mean FMD scores continued to suggest a dose-response relationship between number of depressive episodes and FMD, zero episodes M = 11.0%, 1 episode M = 6.0%, two episodes M = 4.4%, and three episodes M = 2.6%.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSNOTESREFERENCES |
|---|
The relationship between history of depression and FMD held even when current depressive symptoms were statistically controlled. This suggests that history of depression does not continue to influence the vasculature simply or solely via residual depressive symptoms. In fact, both the previously depressed and the never depressed women in our sample showed relatively low CES-D scores. This low level of depressive symptoms may account for our finding that depressive symptoms were not related to FMD, which is in contrast to other reports (20). One might expect that a group of previously depressed women with elevated residual depressive symptoms would show even more attenuated FMD.
We also observed a dose-response effect for the number of depressive episodes on FMD. Increasing depressive episodes was associated with greater endothelial dysfunction. Our results complement reports of impaired FMD in currently depressed participants (18–20). They also complement reports of cardiovascular effects of depression history in middle-aged women (25,26). To our knowledge, prior research has not examined the impact of depression history on endothelium-dependent FMD.
The differences observed in this study are clinically meaningful. Although the exact value of FMD that represents a normal response has not been established (54), a cutoff of >8% has been proposed (55,56). When using this cutoff, participants in this sample who were never depressed showed a normal response, whereas those with even one depressive episode showed an abnormal response.
Furthermore, the increasing decrements in FMD with each successive depressive episode were clinically meaningful. Very small group differences in FMD, on the order of 2%, are predictive of future cardiac events (55,57,58). We observed decreases of similar magnitude with each additional depressive episode. However, our sample size prevents us from concluding whether the differences in FMD between each episode are reproducible. A study is currently under way to establish the reliability of the differences between episodes. Furthermore, given the extra burden involved in collecting number of depressive episodes relative to presence/absence of depression history, this approach cannot be recommended for routine assessment until it is confidently demonstrated that number of depressive episodes yields added predictive power. Moreover, the range of number of episodes was not large in our sample, with the highest being three episodes. Therefore we cannot determine whether and at what point a floor effect might be observed with additional episodes. It is also unknown whether there is a clinical threshold for FMD. Perhaps once FMD reaches a certain degree of impairment, any addition loss of FMD makes little or no difference in prediction of subsequent cardiac events. These questions can only be answered with larger samples and prospective designs.
Implications
Our findings regarding history of depression have implications for determination of appropriate comparison groups in cardiovascular depression research, and suggest that investigating only current depression as a binary phenomenon may be inadequate. First, if one ignores history of depression and only measures current depression, those who are not currently depressed but who have experienced a previous depressive episode may be misclassified into a group thought to be free from the effects of depression. Studies may find a stronger effect for current depression if the comparison group consists of never depressed participants.
Second, measuring only current depression does not account for any residual subclinical depressive symptoms in nondepressed participants with a history of depression. The temporal course of major depression has changed conceptually in recent years from being viewed as a relapsing remitting illness to a chronic illness with varying levels of symptomatology (59). It is known that previously depressed persons have elevated depressive symptoms long after they no longer meet criteria for the disorder (60,61). It is also known that subclinical depressive symptoms are associated with cardiovascular perturbations (62). Studies that do not account for subclinical symptoms in previously depressed persons cannot determine whether effects are due to the previous depressive episode per se, any lingering depressive symptoms, or a combination of the two.
Third, measuring only current depression assumes that those who are currently depressed are experiencing the effects of current depression only. Over the course of the depressive illness (27), patients are thought to experience sensitization to the state of depression, referred to as the "kindling effect" (63–65). Because modal age of first depressive episode is in young adulthood (66), this "kindling" effect can be reasoned backwards, i.e., that a late-middle-aged woman with a major depressive disorder is likely to have had a prior depressive episode or several episodes. To the extent that this is the case, studies that find an effect for current depression, especially in older participants, may actually be capturing the effect of both the current episode and past episodes.
Fourth, measuring only current depression does not account for duration of exposure to depression over the lifespan. Although our findings are limited by sample size, they suggest the importance of accounting for duration of depression over the lifespan.
Limitations and Future Directions
The sample size was small compared with some (20), but not all (19), reports of FMD and depression, possibly leading to Type II errors in detection of differences between groups on covariates. The number of women within each category of depressive episodes was also small, and there was a restricted range in the observed number of depressive episodes (i.e., 0–3). Future studies should employ larger samples with a greater range of number of episodes, and greater representation within each number of depressive episodes. This study employed only postmenopausal women who were primarily white and wealthy, thus limiting its generalizability. Future studies should investigate more diverse samples of men and women. Additionally, the retrospective depression assessment introduces three problems. First, it does not address directionality of the association. It is possible that endothelial dysfunction causes behavior that either is, or presents similar to, depression. Second, an additional unknown and unmeasured variable may account for the association between history of depression and FMD. For these reasons, prospective studies would be ideal, although they require considerable resources, given that onset of first depressive episode tends to occur in the second or third decade of life, menopause several decades later, and incident CHD even later. Third, all self-report and retrospective data are subject to some degree of unreliability due to participant deception, bias, and poor recall. There was no benefit for participants in our study to deceive the interviewer. Bias is most likely a problem when current mood disturbance colors recollection (67,68), and our participant were all free of current depression. Recall was enhanced with the TLFB method. In these ways, we minimized the unavoidable limitations of self-report retrospective data.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSNOTESREFERENCES |
|---|
|
NOTES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSNOTESREFERENCES |
|---|
Supported by grants from the University of Connecticut Center for Interdisciplinary Research in Women’s Health, the American Heart Association, and the University of Connecticut Health Center General Clinical Research Center.
DOI:10.1097/01.psy.0000195868.68122.9e
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSNOTESREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  G. A. Wellenius, K. J. Mukamal, A. Kulshreshtha, S. Asonganyi, and M. A. Mittleman Depressive Symptoms and the Risk of Atherosclerotic Progression Among Patients With Coronary Artery Bypass Grafts Circulation, May 6, 2008; 117(18): 2313 - 2319. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Herbst, R. H. Pietrzak, J. Wagner, W. B. White, and N. M. Petry Lifetime Major Depression is Associated With Coronary Heart Disease in Older Adults: Results From the National Epidemiologic Survey on Alcohol and Related Conditions Psychosom Med, October 1, 2007; 69(8): 729 - 734. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
