Girdler, Susan S. PhD*; Hinderliter, Alan L. MD†; Wells, Ellen C. MD‡; Sherwood, Andrew PhD§; Grewen, Karen M. PhD*; Light, Kathleen C. PhD*
The unexpected results of the Women’s Health Initiative trial,1 which documented an increased incidence of invasive breast cancer associated with postmenopausal hormone use, as well as a significant increase in risk of nonfatal cardiovascular events, have underscored the importance of individualizing hormone therapy. Only one formulation of hormone therapy was examined in the Women’s Health Initiative trial (daily 0.625 mg conjugated equine estrogen plus 2.5 mg medroxyprogesterone acetate), and only women with an intact uterus were studied. Thus, the results of the Women’s Health Initiative trial may not apply to hysterectomized women who do not require the concomitant administration of a progestin or to women who use lower-dose oral estrogen or transdermal estrogen therapies.
Among the many factors to consider in individualizing hormone treatment for postmenopausal women is cigarette smoking. Paradoxically, to the extent that risks associated with hormone therapy are directly related to circulating estrogen levels, then smokers may have a relatively lower risk/benefit ratio, at least for certain types of hormone replacement therapy regimens. There is compelling evidence that smoking is antiestrogenic.2 Observational studies of oral hormone therapy have found that smokers fail to show the significant reduction in risk for hip fracture3 seen in never-smokers who use hormone therapy. Controlled studies of oral hormone therapy report less improvement for smokers versus nonsmokers in measures of bone density,4,5 cholesterol,5 blood pressure (BP), and vascular resistance,6 and these same studies have shown that the smokers have lower levels of circulating estrogens after oral hormone therapy.4–7
The mechanism by which smoking exerts antiestrogenic effects involves smoking-induced alterations in the hepatic metabolism of estrogen, shifting the metabolic pathways of estrogen to inactive metabolites.7,8 We are aware of only 2 studies that have directly compared postmenopausal smokers with nonsmokers receiving different routes of estrogen therapy. In one,9 during oral estrogen plus cyclic progestin therapy, circulating levels of estrogens were 40–70% lower in the smokers than in the nonsmokers, whereas no difference existed during transdermal therapy. In the other study,5 oral hormone therapy was associated with significantly less improvement in plasma lipids and bone mineral density in smokers versus nonsmokers, whereas smoking status did not influence the degree of benefit seen with percutaneous therapy.
Thus, delivery of estrogen directly into systemic circulation, avoiding hepatic first-pass metabolism, may be a more effective route of delivery for postmenopausal smokers. At the same time, because transdermal estrogen is associated with lower levels of circulating estrogens,10,11 and because it delivers natural 17-β serum estradiol (E2), then it may be associated with a better risk/benefit ratio for postmenopausal smokers, a group at high risk of estrogen-deficient disorders of the menopause (eg, osteoporotic fractures)12 as well as cardiac risk factors.13
The purpose of the present study was to compare transdermal with oral estrogen therapy for cardiac risk factors, including BP and total peripheral resistance, in postmenopausal smokers. Additional aims of this study were to examine treatment effects on indices of sympathetic/adrenergic function, including plasma norepinephrine levels as well as cardiac and vascular β-adrenergic receptor responsivity, and endothelial function. We predicted that transdermal estrogen would be associated with greater reductions in sympathetic tone and greater improvement in endothelial function than would oral estrogen therapy in postmenopausal smokers.
MATERIALS AND METHODS
A total of 95 naturally and surgically postmenopausal women, aged 39–72 years, were recruited via newspaper advertisement, provided informed, written consent, and were randomly assigned into protocol. A total of 135 women were initially enrolled into protocol by the study coordinator. Twelve (9%) of these women were excluded owing to low follicle-stimulating hormone (FSH) levels, 10 (7%) were excluded owing to exclusionary medical conditions, and an additional 18 (13%) women withdrew before initial testing and randomization. The study was approved by the Institutional Review Board. Study enrollment occurred between June 1997 and August 1999, with all follow-up testing completed by June 2000.
All women were habitual cigarette smokers, with the requirement that each had smoked a minimum of 10 cigarettes per day for at least 5 years. No woman was attempting to alter her smoking habits during the course of the study, which was confirmed during the monthly compliance checks (see below). All women reported full cessation of menses for at least 9 months before enrollment and no use of any form of hormone replacement or oral contraceptive for at least 12 months. Women with 2 first-degree relatives having breast cancer, women with a history of thrombophlebitis or thromboembolic disorders, gall bladder disease, liver disease, diabetes, cardiovascular disease other than mildly elevated BP (less than 160/95 mm Hg), a history of any endometrial disorder, or women with a history of breast, uterine, or ovarian cancer were excluded. Women reporting the use of any BP, psychotropic, or antilipemic medication or the regular use of herbal medications or alcohol consumption greater than 3 drinks/day were also excluded. Additionally, all women had to have a negative mammogram results and normal Papnicolaou test in the 12 months preceding study enrollment and serum FSH levels greater than 30 IU/mL. To determine whether endometrial hyperplasia or cancer was present, which was exclusionary, the study gynecologist (E.W.) performed transvaginal sonography with measurement of the endometrial thickness during a gynecological examination. Women with endometrial thickness of less than 5 mm were considered normal. Women with an endometrial thickness of 5 mm or greater underwent endometrial biopsy to obtain a tissue diagnosis. Thus, the women enrolled into protocol represented a relatively healthy middle-aged cohort of postmenopausal women. At the same time, the women enrolled into study were largely a sedentary group, with more than 90% of the women in each treatment group reporting no regular exercise.
An overview of the study design is presented in Figure 1. After initial testing, a simple, stratified randomization scheme was used to randomize subjects. Half of the subjects were specifically recruited to come from rural areas, whereas the other half were drawn from urban areas (rural versus urban analyses to be reported elsewhere). Thus, rural versus urban status was used as the stratification variable in the randomization scheme. The allocation sequence was generated by the institution’s Mental Health Clinical Research Center. A senior research staff member who had no other involvement in the study made the group assignments. After approximately two thirds of the subjects had been enrolled and randomized, a decision was made to terminate randomization into the placebo arm of the study when it became apparent that the overall enrollment and completion goals would not be met. However, randomization into the 2 active treatment arms continued as outlined previously. Thus, women were randomly assigned to 1 of 3 conditions: 1) oral conjugated equine estrogen (Premarin; Wyeth-Ayerst, Philadelphia, PA; 0.625 mg/d), n = 36; 2) transdermal 17β-E2 (Climara; Berlex Laboratories, Wayne, NJ; 0.05 mg/d), n = 35; and 3) placebo, n = 24. To control for any nonspecific effects associated with taking oral medication versus wearing a transdermal system, half the women received oral estrogen placebo and half received transdermal estrogen placebo. Although, clinically, the concomitant use of a progestin with estrogen therapy is indicated only in women with an intact uterus, because progesterone has been shown to antagonize estrogen’s actions,14 to avoid any potential confounding of hysterectomy status with progestin therapy in the present study, all women in the active treatment arms also received oral, medroxyprogesterone acetate (Provera; Pharmacia & Upjohn, Peapack, NJ; 2.5 mg/d), while the placebo group received a daily placebo capsule.
Of the 95 women randomized, 7 (7%) withdrew or were excluded before completing the full 6-month treatment and retesting protocol (3 assigned to transdermal, 2 assigned to oral, and 2 assigned to the placebo arm). Reasons for drop out included lack of adherence to the protocol (n = 3), illness (n = 1), lack of interest (n = 1), symptoms (n = 1 who was randomized to placebo), and development of a benign breast lump (n = 1). Consequently, 88 women completed the full 6-month protocol. However, 6 of these women were dropped from the final analyses (3 from the oral and 3 from the transdermal group) based on posttreatment serum E2 levels indicative of noncompliance (ie, less than 10% increase from pre- to posttreatment or a decrease from pre- to posttreatment in serum E2 concentrations). Thus, the remaining 82 women who make up this study completed the full 6 months of testing with satisfactory adherence, as determined both by monthly pill/patch count and plasma E2 changes. Total time of study enrollment, including medical screenings and testing, varied between 7 and 10 months.
The institution’s General Clinical Research Center provided masking of the active and placebo preparations used in the present study. Only the General Clinical Research Center and study gynecologist (E.W.) had access to the participant treatment codes. Thus, both the participants and the research staff having contact with them were initially blind to their treatment condition. However, all women with no prior hysterectomy who were randomly assigned to active treatment arms became unblinded with the onset of vaginal bleeding. The proportion of subjects in the oral versus transdermal estrogen conditions who became unblinded did not differ. All research staff and investigators involved in data editing and analyses remained blinded to treatment code. Each subject was tested twice, once before and once after 6 months of treatment. Between test sessions, subjects visited the laboratory on a monthly basis for assessment of compliance (via pill and patch counts and review of a daily medication log), BP levels, and side effects. Testing was identical each time and consisted of a 4-hour laboratory testing protocol (see below). All testing sessions were conducted between 8:00 am and 12:00 pm. Subjects were instructed to refrain from consuming caffeine and all over-the-counter medications on the day of testing. However, subjects were allowed to smoke ad libitum before arrival at the laboratory to minimize nicotine deprivation levels and withdrawal symptoms.
BPs were recorded noninvasively by using the auscultatory technique. During testing, a laboratory-built, semiautomated BP monitor was used to operate the BP cuff, displaying cuff pressure and Korotkoff sounds on a computer screen with the Videograph computer system (PAS, Raleigh, NC). Systolic BP corresponded to the onset of Korotkoff sounds and diastolic BP corresponded to the disappearance of Korotkoff sounds. Before beginning the protocol, 3 to 4 manual stethoscopic readings were taken initially with a sphygmomanometer to insure correct placement of the microphone.
Impedance cardiography was used to permit noninvasive monitoring of cardiac performance,15 including stroke volume (SV), heart rate, and pre-ejection period. A custom-designed impedance cardiograph (HIC-100, Model 100; Bio-Impedance Technology, Inc, Chapel Hill, NC) was used in conjunction with a tetrapolar band electrode configuration to record impedance dZ/dt and Zo signals. Impedance and electrocardiogram signals were processed online by specialized computer software (BIT, Chapel Hill, NC) with subsequent manual editing to improve accuracy. For each minute of interest, a 30-second continuous sample of waveforms (obtained concurrently with BP) was processed to generate an ensemble-averaged cardiac cycle, from which SV was determined by means of the Kubicek et al16 equation, and heart rate was determined by the mean interbeat interval. Cardiac output and total peripheral resistance for these same minutes were then calculated with standard formulas.15
After instrumentation for cardiovascular monitoring, each subject was seated in a comfortable chair, an intravenous line was established in an arm vein, and a curtain was drawn to prevent the subject from viewing the intravenous line. A minimum of 15 minutes elapsed between establishing the intravenous line and beginning baseline rest. Subjects were exposed to the following conditions with a 5-minute recovery period following each stressor:
Subjects rested quietly for 10 minutes. Cardiovascular measures were collected simultaneously during minutes 1, 3, 5, 8, and 10. Blood was sampled for baseline levels of norepinephrine during minute 10. This blood draw also provided the E2, estrone, and estriol samples.
Each subject was presented with a hypothetical situation involving an interpersonal hassle and was given 2 minutes in which to prepare (Speech Preparation) to give a 3-minute, tape-recorded talk (Speech) describing what her actions and emotional responses would be in the situation. Cardiovascular measures were taken during minute 2 of Speech Preparation and during minutes 1 and 3 of the Speech. Blood was sampled for norepinephrine at the end of minute 1 and minute 2 of the Speech.
This test is composed of 3 subsets. Each subject first read a chart of colored squares, then read aloud a list of color words printed in black ink, and finally scanned a list of color words printed in noncorresponding ink (eg, the word “blue” printed in red ink), stating aloud the color of the ink in which the color word was printed. Cardiovascular measures were recorded during minute 1 of the third component only, and plasma norepinephrine was sampled 45 seconds after the initiation of the third component.
Forehead cold pressor.
The experimenter held a plastic bag filled with ice and water (4–6°C) on the forehead for 2 minutes. Cardiovascular measures were averaged across minutes 1 and 2.
Mental stress testing was followed by a 20-minute recovery period, during which time the subject rested quietly in the supine position. BP was measured continuously using the Finapres (Ohmeda, Madison, WI) noninvasive BP monitor. This instrument, which uses the vascular unloading technique to measure systolic, diastolic, and mean BP on a beat-by-beat basis, has been validated against intraarterial measures under various conditions, including pressure responses to phenylephrine.17
The standardized isoproterenol sensitivity test was used to evaluate β-adrenergic receptor responsiveness in terms of the chronotropic dose of isoproterenol required to increase heart rate by 25 beats/min (CD25).18 Progressively increasing bolus doses of isoproterenol (0.125, 0.25, 0.5, 1.0, 2.0, and 4.0 μg) were injected into an arm vein until an increase in heart rate of at least 25 beats/min was observed. Heart rate responses following each dose were computed as the shortest of 3 successive electrocardiogram R-R intervals following drug injection, compared to the shortest 3 R-R intervals at rest (preinjection). Following each dose, the next higher dose was not injected for at least 5 minutes or until cardiovascular activity had returned to resting levels, usually within 5 to 10 minutes. The linear regression model of log dose/heart rate response for each subject was used to determine CD25 exactly by interpolation. The CD25 measure provides an index of cardiac β1-receptor responsiveness.19 A vascular β2-receptor responsiveness index was also derived by determining the vasodilatory isoproterenol dose required to decrease total peripheral resistance by 40% (VD40), by using log dose/total peripheral resistance response interpolation.20 Both the CD25 and VD40 indices are inversely related to receptor responsiveness.
Brachial artery reactivity was assessed by measuring changes in arterial diameter induced by reactive hyperemia and nitroglycerin. Ultrasound images of the right brachial artery proximal to the antecubital fossa were acquired with a 7.5-mHz linear-array transducer. Baseline images were obtained after 10 minutes of supine rest. Flow-mediated (endothelium-dependent) dilation was assessed by determining the change in arterial diameter in response to reactive hyperemia. Reactive hyperemia was induced by inflating a pneumatic occlusion cuff placed around the forearm to a suprasystolic pressure (approximately 200 mm Hg) for 5 minutes. Images of the artery were recorded for 2 minutes after cuff deflation. After 10 minutes of rest, a second baseline image was acquired. Sublingual nitroglycerin spray (0.4 mg) was then administered to determine endothelium-independent vasodilation, and ultrasound images were acquired for the subsequent 5 minutes.
Brachial artery images were recorded on VHS video tapes for subsequent analysis. Measurements were performed on a personal computer equipped with a video frame grabber and customized software (Brachial Tools, Medical Imaging Applications, LLC, Coralville, IA). Arterial diameter was measured from the media-adventitial interfaces of the proximal and distal arterial walls. Data from at least 3 consecutive end-diastolic frames were averaged for each baseline measurement, and for at least 3 frames near the maximum dilation during reactive hyperemia and after administration of nitroglycerin.
Flow-mediated and nitroglycerin-mediated dilation were calculated as percentage changes in diameter from baseline.
Plasma levels of norepinephrine were determined by using the high-performance liquid chromatography technique. The lower limit of quantification with this system is 25 pg/mL, and the intra- and interday coefficients of variation are less than 10%. Blood for estrogens was collected into serum separator tubes and allowed to clot for 2 hours before centrifuging and pipetting serum into aliquots, which were then immediately frozen at −80°C until assay. Serum levels of E2, estron, and estriol were determined by using RIA kits from Diagnostic Systems Laboratories, Inc (Webster, TX). The sensitivity of the E2 assay is 2.2 pg/mL, and the specificity of the antiserum for E2 is high, showing only 0.01–0.2% cross-reactivity with most other steroid hormones, with the exception of estrone, D-equilenin, and 17β-estardiol-3-glucuronide, for which there is 2–3.4% cross-reactivity. The sensitivity of estrone assay is 1.2 pg/mL, and its specificity is high, showing 0–0.4% cross-reactivity with other steroid hormones, with the exception of estrone sulfate, for which there is 2.02% cross-reactivity, and 17β-E2, for which there is 1.25% cross-reactivity. The assay for estriol demonstrates a sensitivity of 0.03 ng/mL and minimal cross-reactivity with most other steroid hormones (0–0.74%), with the exception of 16-epiestriol for which there is 3.7% cross-reactivity.
Initial power calculations to determine sample size were based on the primary measures central to our hypotheses relating to route of estrogen therapy and vascular sympathetic tone. Thus, estimates were based on the ability to detect group differences in change from pre- to posttreatment in β-adrenoceptor responsivity and total peripheral resistance. Previous21 and ongoing studies from our laboratory provided the data used to calculate expected differences in β-adrenoceptor responsivity and vascular resistance (ie, 0.15 μg and 150 total peripheral resistance units, respectively). These estimates revealed that a sample size of 25 per group would yield 90% power to detect a difference in means of 0.15 μg for β-adrenoceptor responsivity, whereas sample sizes of 33 per group would yield 80% power to detect expected total peripheral resistance differences. Thus, although sample size goals were met based on initial enrollment and randomization, the loss of 16% of the sample to dropout or noncompliance has implications for type II error rates in the present study.
Our first analytic strategy was to examine whether groups differed in any demographic factor or in their pretreatment responses to the laboratory protocol by using either one-way analysis of variance (ANOVA) or χ2 analyses. For each cardiovascular variable at pretreatment, a 3(Treatment) × 5(Condition: Baseline, Speech Prep, Speech, Stroop, Cold Pressor) repeated-measures ANOVA was employed, with Condition as the repeated factor. For plasma norepinephrine a 3(Treatment) × 4(Condition: Baseline, minute 1 Speech, minute 2 Speech, Stroop) repeated-measures ANOVA was used. Plasma estrogen levels and the ratio of estrogens were examined by using a 3(Treatment) × 2(Time: Pre- and Posttreatment) repeated-measures ANOVA.
Treatment-related changes in cardiovascular and neuroendocrine measures were analyzed by examining within-subject changes from pretreatment to posttreatment. Thus, for each dependent measure, a change score was created (posttreatment − pretreatment) and analyzed by using Treatment × Condition repeated-measures ANOVAs. Where significant Treatment effects emerged, examination of least-squares means or the use of within-subjects paired t tests was used to identify the source of the effect. Owing to the fact that 5 subjects refused the cold pressor test during their posttreatment session, responses to the cold pressor were analyzed separately in the subset of 77 women for whom both pre- and posttreatment values were available. Problems with the intravenous line or inability to obtain all blood samples yielded 53 women for analyses of plasma norepinephrine (21 transdermal, 21 oral, and 11 placebo), and 79 women for analyses of β-receptor responsivity (29 transdermal, 29 oral, and 21 placebo). Technically adequate vascular ultrasound studies were acquired before and after therapy in 55 subjects (20 transdermal, 25 oral, and 10 placebo).
As summarized in Table 1, there were no significant differences in any subject demographic factor, including pretreatment levels of mean arterial pressure (MAP), age, body mass index, pack-years of cigarette smoking, years postmenopausal, ethnic breakdown, or percentage with mild systolic hypertension. Although a smaller percentage of hysterectomized women were randomized to placebo, this proportional difference was not statistically significant (χ2 = 4.2, P = .12). Despite the nonsignificant results, owing to the smaller percentage of hysterectomized women randomized to placebo, analyses of our primary dependent measures, total peripheral resistance, and norepinephrine were repeated by using hysterectomy status (yes/no) as a covariate. Of those hysterectomized, 5 (31%) in the transdermal, 2 (14%) in the oral, and 1 (20%) in the placebo group had undergone double oophorectomy, while the remaining hysterectomized women had both ovaries in tact. Groups did not differ in the proportion oophorectomized (χ2 = 1.2, P > .50). Additionally, groups did not differ in any cardiovascular measure assessed in the laboratory at pretreatment, either during baseline rest or during any of the stressors (all P values > .33). Similarly, there were no group differences in either baseline or stress levels of plasma norepinephrine or in flow-mediated or nitroglycerin-mediated dilation of the brachial artery at pretreatment testing (all P values > .55). Thus, before treatment, there were no appreciable differences between the groups in any demographic factor or in their cardiovascular and neuroendocrine responses to the stressors.
At pretreatment, there were no group differences in any of the serum estrogen levels or in their ratios, with all women demonstrating expected postmenopausal levels (Table 2). Only the 2 treatment groups exhibited a significant increase in their E2 levels at 6 months (Time × treatment: F[2, 79] = 29.4, P < .001). However, although both treated groups had higher E2 levels at 6 months relative to the placebo group (P values < .001), the oral estrogen group had nearly twice the level of serum E2 relative to the transdermal group at posttreatment (P < .001). Also, as expected, the oral estrogen group showed a substantial and significant increase in serum estrone levels after treatment (Time × treatment: F[2,79] = 38.3, P < .001), exhibiting posttreatment levels greater than either the transdermal or placebo groups (P values < .001), who did not differ from each other. For serum estriol levels, an unexpected main effect of Time was obtained (F[1,80] = 7.8, P < .01) because all 3 groups showed an increase in serum estriol from pre- to posttreatment. However, subsequent within-subjects, paired t tests revealed that this increase was statistically significant in the oral group only (P < .05).
Regarding the estriol/estrone ratio, a significant Time × Treatment interaction was obtained (F[2,79] = 10.1, P < .001), because the oral estrogen group showed a significantly lower ratio at posttreatment than either of the other groups (P values < .001), indicating that the vast majority of the excess estrone in the oral group was not converted to the biologically active estriol, but instead to inactive metabolites. For the E2/estrone ratio at posttreatment, the transdermal group exhibited a significantly greater ratio than either of the other groups (P values < .001), who did not differ from each other. Thus, the transdermal group exhibited an E2/estrone ratio closer to that seen in premenopausal women (≈ 1).
Figure 2 depicts change in total peripheral resistance from pre- to posttreatment during baseline rest, as well as during stressors. For total peripheral resistance, a significant main effect of Treatment Group was obtained during rest and behavioral stressors (F[2,79] = 4.0, P < .05), and a similar effect was observed during the cold pressor test (F[2,73] = 2.9, P = .06). Examination of least-squares means indicated that, although both treatment groups showed greater reductions from pre- to posttreatment in total peripheral resistance compared with the placebo group during Speech Preparation and cold pressor stress (P values < .05), only the transdermal group showed greater reductions in total peripheral resistance than the placebo group during baseline rest and speech stress (P values < .05). The results were virtually unchanged when analyses were rerun with hysterectomy status as a covariate (main effect of treatment: F[2,78] = 2.96, P = .05).
Similar main effects of Treatment Group were obtained for cardiac output (F[2.79] = 4.4, P = .01 and F[2,75] = 4.8, P = .01) and SV (F[2.79] = 2.7, P = .07 and F[2,75] = 3.0, P = .05), because, relative to the placebo group, both treated groups showed treatment-related increases in cardiac output at baseline and during stressors, although these effects were more consistent and pronounced in the transdermal group (Table 3). Moreover, only the transdermal group showed treatment-related increases in SV at rest (P < .05) and during Speech Preparation (P < .01) and cold pressor stress (P < .05). Regarding MAP, although there was no overall main effect of Treatment Group, based on a priori hypotheses, within-subjects, paired comparisons t tests indicated that only the transdermal group demonstrated treatment-related reductions in their BP during baseline rest and behavioral stressors (P values < .05; Table 3). There were no significant treatment-related effects or group differences for heart rate.
Figure 3 depicts treatment-related change in plasma norepinephrine during baseline rest and mental stressors. A significant main effect of Treatment Group was observed for plasma norepinephrine (F[2,51] = 3.5, P < .05) since only the transdermal group showed greater reductions in norepinephrine relative to the placebo group during all behavioral stressors (all P values < .05). These results were virtually identical when analyses were rerun with hysterectomy status as a covariate (main effect of treatment: F[2,49] = 4.24, P < .05).
There was a trend for a treatment-related change in vascular β2-receptor responsivity (main effect of treatment: F[2,78] = 2.6, P = .07; Figure 4). Based on a priori hypotheses, subsequent examination of least-squares means revealed that this effect was driven by the treatment-related decrease in vasodilatory isoproterenol dose required to decrease total peripheral resistance by 40% (VD40) seen in the transdermal group relative to the placebo group (P < .05), whereas the oral estrogen group did not differ from placebo. Thus, when compared with placebo-treated women, only transdermal estrogen therapy tended to decrease the vasodilatory dose of isoproterenol (ie, tended to increase vascular β2-receptor responsivity). There were no statistically significant effects involving β1-adrenoceptor responsivity (ie, CD25).
The treatment-related changes in brachial artery reactivity are illustrated in Figure 5. The mean values for flow-mediated and nitroglycerin-mediated dilation at baseline were 3.5% ± 3.2% and 16.5% ± 5.9%, respectively. Although the effect of Treatment Group on flow-mediated dilation was not statistically significant (main effect of treatment: F[2,52] = 1.85, P = .17), an increase from baseline was observed in subjects treated with transdermal estrogen (P = .03) but not for subjects randomly assigned to therapy with oral estrogen or placebo. No clear trend was observed in the relation of nitroglycerin-mediated dilation to Treatment Group (main effect of treatment: F[2,46] = 0.19, P = .75).
In this study, designed to compare different routes of estrogen therapy in postmenopausal smokers for measures relating to cardiac risk, our predictions that 6 months of transdermal versus oral estrogen therapy would be associated with greater reductions in measures reflecting vascular sympathetic tone were largely supported. Specifically, when compared with the placebo group, we observed that women assigned to transdermal estrogen showed more consistent reductions in vascular resistance at rest and in response to mental stress than women assigned to oral estrogen, and only the transdermal group showed treatment-related reductions in stress levels of plasma norepinephrine, in resting and stress MAP levels, in increases in vascular β2-receptor responsivity, and in endothelium-dependent vasodilation.
The clinical significance of these results is based on the tenet that excessive cardiovascular reactivity to stress has a pathophysiological role in neurogenic hypertension.22 Historically, studies of the so-called “reactivity hypothesis” have used the cold pressor test and have shown that BP responses to stress in young, healthy individuals predict the development of hypertension up to 36 years later.23,24 More recently, numerous studies using mental stressors such as speech have now shown that BP and cardiac responses to these laboratory stressors in young healthy individuals predict future BP levels and hypertensive status.22,25–27 Evidence that responses to standardized mental stressors also predict cardiac events comes from studies in untreated hypertensive patients showing that high BP reactors are more than twice as likely to experience a myocardial infarction over a 14-year follow-up period28 and from studies in coronary patients showing that mental stress-induced ischemia is associated with a 3-fold increase in risk of clinical events (eg, cardiac death, myocardial infarction, angioplasty), even after controlling for medical risk factors.20,29,30 The validity of the reactivity model and the use of surrogate markers to predict cardiac risk is further bolstered by prospective studies showing that increased total peripheral resistance, as well as BP responses to mental stress, predict increased left ventricular mass and decreased cardiac function over a 2-year period.31 Thus, increased vascular tone that leads to increased total peripheral resistance appears to play a primary role in cardiac remodeling and function,31 and the Framingham studies have documented that left ventricular hypertrophy is the strongest predictor of cardiovascular mortality other than age.32
The more consistent treatment-related reductions in total peripheral resistance seen in the postmenopausal smokers treated with transdermal estrogen are noteworthy. Both in vitro and in vivo studies have provided evidence that estrogen may enhance vasodilatation through multiple routes, including endothelium-dependent mechanisms.33 Endothelial function is known to be impaired in smokers.34,35 Previous investigators, however, have described enhanced endothelium-dependent vasodilation in postmenopausal women treated with estrogen.36,37 Thus, greater improvement of endothelial function seen with transdermal versus oral estrogen therapy may be one factor contributing to the reductions in total peripheral resistance that we observed with transdermal estrogen treatment.
An additional mechanism that may contribute to the robust reductions in vascular tone seen with transdermal therapy relates to our observation of greater β2-adrenoceptor responsivity seen only in the transdermal group compared with placebo-treated women following therapy. Treatment-related increases in β2-receptor responsivity may have particular relevance for smokers because studies have documented a down-regulation of β-adrenergic receptors in smokers.38,39 Studies in rats have previously confirmed that estrogen increases β-adrenoceptor–mediated relaxation of mesenteric arteries,40 while studies in humans have documented greater density and responsivity of lymphocyte β2-receptors in premenopausal women tested during high-hormone, versus low-hormone, phases of their menstrual cycle.41 Enhanced function of the vascular endothelium may also contribute to β2-receptor responsivity. Dawes et al42 demonstrated that β-adrenergic vasodilator responses in the human forearm are dependent on nitric oxide synthesis by the endothelium. Moreover, individuals with impaired endothelium-dependent arterial dilation exhibit greater total peripheral resistance responses during laboratory stressors than individuals with greater endothelium-dependent dilation.43
Another direct mechanism by which transdermal estrogen increases vascular β2-adrenoceptor responsivity may be through associated decreases in circulating catecholamines. Although others have documented reductions in plasma norepinephrine with oral hormone therapy in predominantly nonsmoking women,44,45 this is the first study to document hormone therapy–related reductions in plasma catecholamine levels in smokers. Again, treatment-related reductions in circulating catecholamine levels may be particularly relevant for smokers because they have been shown to exhibit chronically elevated levels compared with nonsmokers.46,47 Evidence for regulation of peripheral β-adrenoceptors by catecholamines in humans comes from studies showing inverse relationships between circulating catecholamine levels and lymphocyte β-adrenoceptor density or responsivity.48,49 Thus, our findings suggest that a 6-month trial of hormone therapy with transdermal estrogen in postmenopausal smokers may decrease vascular tone, in part, via reductions in plasma norepinephrine levels and subsequent enhancement of vascular β2-adrenergic receptor responsivity.
Although we also observed treatment-related increases in basal and stress-induced stroke volume and cardiac output, which were especially apparent in the transdermal group, these increases may be most parsimoniously explained as a result of the reductions in total peripheral resistance and afterload on the heart and do not necessarily indicate any increase in cardiac contractility with the hormone therapy regimens. This explanation is further supported by our lack of finding for any significant treatment-related effect on cardiac β1-adrenoceptor responsivity. Although we failed to observe any overall treatment effects for MAP, when compared with their own treatment levels, the transdermal group was the only group to show significant decreases in both resting and stress-induced BP. Given that cardiac output increased with treatment, especially in the transdermal group, reductions in their MAP levels were likely mediated, at least in part, by their treatment-related reductions in total peripheral resistance.
In conclusion, our results indicate a benefit of short-term hormone therapy, using transdermal estrogen, in postmenopausal smokers for certain cardiovascular and adrenergic measures, particularly vascular tone, MAP, and circulating catecholamines. In contrast to our prior research in women who were predominantly nonsmokers,45 oral conjugated equine estrogens had minimal benefits in these smokers. In light of the recent results from the Women’s Health Initiative, the clinical implications of our findings must be interpreted cautiously until longer-term, controlled clinical trials involving transdermal estrogen are conducted. An additional caveat to bear in mind is that the women in the present study were, on average, at least 10 years younger than the Women’s Health Initiative cohort. Thus, it is possible that we would not have seen the same degree of benefit with transdermal estrogen in older postmenopausal smokers or that either of the active treatment arms would have been associated with deleterious changes in cardiovascular indices in older women. Accordingly, the present results are intended to have implications for short-term hormone therapy in young, healthy, postmenopausal smokers.
Another limitation to this initial study comparing route of estrogen therapy and cardiovascular indices in postmenopausal smokers is the relatively small sample size used, rendering our results as preliminary. Our small cell sizes and overall loss of approximately 30% of norepinephrine and endothelial data owing to technical difficulties also implicate reduced statistical power as a possible explanation for fewer significant effects involving oral estrogen therapy. It should be noted, however, that proportionately less norepinephrine and endothelial data loss came from the oral estrogen group (3 and 16%) than from both the transdermal (6 and 35%) and placebo (48 and 52%) groups. Thus, it is unlikely that data loss biased us against observing significant decreases in plasma norepinephrine and increases in endothelium-dependent vasodilation with oral estrogen, effects that were seen in the transdermal group.
Despite the study limitations, the consistency of the results seen for transdermal estrogen therapy, across multiple measures reflecting vascular sympathetic tone, in combination with the overall support obtained for our a priori hypotheses that transdermal estrogen would be associated with greater beneficial changes in smokers, provides the evidence needed to support a larger, randomized trial, using an intent-to-treat statistical approach. Such a design would allow for definitive findings to guide physicians in choosing hormone therapy strategies for postmenopausal smokers. Given that the primary clinical indication for estrogen therapy (with or without a progestin) is for the short-term management of menopausal, estrogen-deficiency symptoms, then our results suggest that short-term transdermal estrogen may be a safe treatment option in relatively young, healthy, postmenopausal smokers. In addition to the benefits our study observed, the transdermal delivery of physiologic 17β-E2, combined with lower circulating levels of E2 and estrone, as well as more physiologic relationships between the active estrogens, may all combine to impart lesser risk for adverse effects of hormone therapy. Clearly, longer-term studies of different hormone replacement regimens, powered to test for an effect of smoking status, are needed to effectively individualize hormone therapy for all women.
1. Writing Group for the Women’s Health Initiative Investigators. Risks and benefits of estrogen plus progestin in healthy postmenopausal women. JAMA 2002;288:321–33.
2. Baron JA, La Vecchia C, Levi F. The antiestrogenic effect of cigarette smoking in women. Am J Obstet Gynecol 1990;162:502–14.
3. Kiel DP, Baron JA, Anderson JJ, Hannan MT, Felson DT. Smoking eliminates the protective effect of oral estrogens on the risk for hip fracture among women. Ann Intern Med 1992;116:716–21.
4. Jensen J, Christiansen C, Rodbro P. Cigarette smoking, serum estrogens and bone loss during hormone-replacement therapy early after menopause. N Engl J Med 1985;313:973–5.
5. Jensen J, Christianson C. Effects of smoking on serum lipoproteins and bone mineral content during postmenopausal hormone replacement therapy. Am J Obstet Gynecol 1988;159:820–5.
6. Girdler SS, Hinderliter AL, West SG, Grewen K, Steege J, Light KC. Postmenopausal smokers show reduced hemodynamic benefit from oral hormone replacement. Am J Cardiol 2000;86:590–2.
7. Cassidenti DL, Vijod AG, Vijod MA, Stanczyk FZ, Lobo RA. Short-term effects of smoking on the pharmacolinetic profiles of micronized estradiol on postmenopausal women. Am J Obstet Gynecol 1990;163:1953–69.
8. Michnovic JJ, Hershcopf RJ, Naganuma H, Bradlow HL, Fishman J. Increased 2-hydroxylation of estradiol as a possible mechanism for the anti-estrogenic effect of cigarette smoking. N Engl J Med 1986;315:1305–9.
9. Geisler J, Omsjo IH, Helle SI, Ekse D, Silsand T, Lonning PE. Plasma oestrogen fractions in postmenopusal women receiving hormone replacement therapy: influence of route of administration and cigarette smoking. J Endocrinol 1999;162:265–70.
10. Corson SL. A decade of experience with transdermal estrogen replacement therapy: overview of key pharmacologic and clinical findings. Int J Fertil 1993;38:79–91.
11. Powers MS, Schenke L, Darley PE, Good WR, Balestra JC, Place VA. Pharmacokinetics and pharmacodynamics of transdermal dosage forms of 17β-estradiol: comparison with conventional oral estrogens used for hormone replacement. Am J Obstet Gynecol 1985;152:1099–106.
12. Williams AR, Weiss NS, Ure CL, Ballard J, Daling JR. Effect of weight, smoking, and estrogen use on the risk of hip and forearm fractures in postmenopausal women. Obstet Gynecol 1982;60:695–9.
13. Jacobson BH, Aldana SG, Adams TB, Quirk M. The relationship between smoking, cholesterol, and HDL-C levels. Women Health 1995;23:27–38.
14. Graham JD, Clarke CL. Physiological action of progesterone in target tissues. Endocr Rev 1997;18:502–19
15. Sherwood A, Allen MT, Fahrenberg J, Kelssey RM, Lovallo WR, van Doornen LJP. Committee report: methodological guidelines for impedance cardiography. Psychophysiology 1990;27:1–23.
16. Kubicek WG, Karnegis JN, Patterson RP, Witsoe DA, Mattson RH. Development and evaluation of an impedance cardiac output system. Aerosp Med 1966;37:1208–12.
17. Parati G, Casadei R, Gropelli A, Di Rienzo M, Mancia G. Comparison of finger and intra-arterial blood pressure monitoring at rest and during laboratory testing. Hypertension 1989;13:647–55.
18. Cleaveland CR, Rangno RE, Shand DG. A standardized isoproterenol sensitivity test. Arch Intern Med 1972;130:47–52.
19. Dimsdale J, Ziegler M, Graham R. The effect of hypertension, sodium, and race on isoproterenol sensitivity. Clin Exp Hypertens A 1988;10:747–56.
20. Sherwood A, Hinderliter AL. Responsiveness of alpha- and beta-adrenergic receptor agonists. Am J Hypertens 1993;6:630–5.
21. Girdler SS, Jamner LD, Jarvik M, Soles Jr, Shapiro D. Smoking status and nicotine administration differentially modify hemodynamic stress reactivity in men and women. Psychosom Med 1997;59:294–306.
22. Matthews KA, Woodall KL, Allen MT. Cardiovascular reactivity to stress predicts future blood pressure status. Hypertension 1993;22:479–85.
23. Menkes MS, Matthews KA, Krantz DS, Lundberg U, Mead LA, Qaqish B, et al. Cardiovascular reactivity to the cold pressor test as a predictor of hypertension. Hypertension 1989;14:524–30.
24. Kasagi F, Akahoshi M, Shimaoka K. Relation between cold pressor test and development of hypertension based on 28-year follow-up. Hypertension 1995;25:71–6.
25. Girdler SS, Hinderliter AL, Brownley KA, Turner JR, Sherwood A, Light KC. The ability of active versus passive coping tasks to predict future blood pressure levels in normotensive men and women. Int J Behav Med 1996;3:233–50.
26. Markovitz JH, Raczynski JM, Wallace D, Chettur V, Chesney MA. Cardiovascular reactivity to video game predicts subsequent blood pressure increases in young men: the CARDIA study. Psychosom Med 1998;60:186–91.
27. Light KC, Girdler SS, Sherwood A, Bragdon EE, Brownley KA, West SG, et al. High stress responsivity predicts later blood pressure only in combination with positive family history and high life stress. Hypertension 1999;33:1458–64.
28. Alderman MH, Ooi WL, Madhavan S, Cohen H. Blood pressure reactivity predicts myocardial infarction among treated hypertensive patients. J Clin Epidemiol 1990;43:859–66.
29. Jiang W, Babyak M, Krantz DS, Waugh RA, Coleman RE, Hanson MM, et al. Mental stress-induced myocardial ischemia and cardiac events. JAMA 1996;275:1651–6.
30. Sheps DS, McMahon RP, Becker L, Carney RM, Freedland KE, Cohen JD, et al. Mental stress-induced ischemia and all-cause mortality in patients with coronary artery disease. Circulation 2002;105:1780–4.
31. Kapuku GK, Treiber FA, Davis HC, Harshfield GA, Cook BB, Mensah GA. Hemodynamic function at rest, during acute stress, and in the field. Hypertension 1999;34:1026–31.
32. Levy D, Garrison RJ, Savage DD, Kannel WB, Castelli WP. Prognostic implications of echocardiographically determined left ventricular mass in the Framingham Heart Study. N Engl J Med 1990;322:1561–6.
33. Lieberman EH, Gerhard MD, Uehata A, Walsh BW, Selwyn AP, Ganz P, et al. Estrogen improves endothelium-dependent, flow-mediated vasodilation in postmenopausal women. Ann Intern Med 1994;121:936–41.
34. Lekakis J, Papamichael C, Vemmos C, Nanas J, Kontoyannis D, Stamatelopoulos S, et al. Effect of acute cigarette smoking on endothelium-dependent brachial artery dilatation in healthy individuals. Am J Cardiol 1997;79:529–31.
35. Celermajer DS, Sorensen KE, Georgakopoulos D, Bull C, Thomas O, Robinson J, et al. Cigarette smoking is associated with dose-related and potentially reversible impairment of endothelium-dependent dilation in healthy young adults. Circulation 1993;88:2149–55.
36. Sudhir K, Jennings GL, Funder JW, Komesaroff PA. Estrogen enhances basal nitric oxide release in the forearm vasculature in perimenopausal women. Hypertension 1996;28:330–4.
37. Lang U, Baker R, Clarke K. Estrogen-induced increases in coronary blood flow are antagonized by inhibitors of nitric oxide synthesis. Eur J Obstet Gynecol Reprod Biol 1997;74:229–35.
38. Laustiola KE, Lassila R, Kaprio J, Koskenvuo M. Decreased β-adrenergic receptor density and catecholamine response in male cigarette smokers: a study of monozygotic twin pairs discordant for smoking. Circulation 1988;78:1234–40.
39. Hjemdahl P, Larsson K, Johansson MC, Zetterlund A, Eklund A. β-adrenoceptors in human alveolar macrophages isolated by elutriation. Br J Clin Pharmacol 1990;30:673–82.
40. Ferrer M, Meyer M, Osol G. Estrogen replacement increases β–adrenoceptor-mediated relaxation of rat mesenteric arteries. J Vasc Res 1996;33:124–31.
41. Wheeldon NM, Newnham DM, Coutie WJ, Peters JA, McDevitt DG, Lipworth BJ. Infleunce of sex-steroid hormones on the regulation of lymphocyte x-adrenoceptors during the menstrual cycle. Br J Clin Pharmacol 1994;37:583–8.
42. Dawes M, Chowienczyk PH, Riter JM. Effects of inhibition of the l-arginine/nitric oxide pathway on vasodilation caused by β-adrenergic agonists in human forearm. Circulation 1997;95:2293–7.
43. Sherwood A, Johnson K, Blumenthal JA, Hinderliter AL. Endothelial function and hemodynamic responses during mental stress. Psychosom Med 1999;61:365–70.
44. Blum I, Vered Y, Lifshitz A, Harel D, Blum M, Nordenberg Y, et al. The effect of estrogen replacement therapy on plasma serotonin and catecholamines of postmenopausal women. Isr J Med Sci 1996;32:1158–62.
45. Light KC, Hinderliter AL, West SG, Grewen KM, Steege JF, Sherwood A, et al. Hormone replacement improves hemodynamic profile and left ventricular geometry in hypertensive and normotensive postmenopausal women. J Hypertens 2001;19:269–78.
46. Grassi G, Seravalle G, Calhoun DA, Bolla G, Mancia G. Cigarette smoking and the adrenergic nervous system. Clin Exp Hypertens A 1992;14:251–60.
47. Pomerleau OF. Nicotine and the central nervous system: biobehavioral effects of cigarette smoking. Am J Med 1992;93:2S–7S.
48. Davies AO, Lefkowitz RJ. Regulation of adrenergic receptors. In: Lefkowitz RJ, editor. Receptors and recognition. New York (NY): Chapman and Hall; 1981. p. 83–121.
49. Frey MJ, Ecker H, Wilson JR, Molinoff PB. Regulation of β-adrenergic receptors on circulating lymphocytes and cardiac muscle. J Cardiovasc Pharmacol 1987;10:545–9.