The Effect of Perimenopausal Transdermal Estradiol and Micronized Progesterone on Markers of Risk for Arterial Disease

Abstract

Background

The arterial effects of hormone therapy remain controversial. This study tested the effects of transdermal estradiol plus intermittent micronized progesterone (TE + IMP) in healthy perimenopausal and early postmenopausal women on several mechanisms involved in the pathophysiology of arterial disease.

Methods

Healthy perimenopausal and early postmenopausal women, ages 45 to 60 years, were enrolled in this randomized, double-blind, placebo-controlled trial. Women were randomized to receive TE (0.1 mg/day) + IMP (200 mg/day for 12 days) or identical placebo patches and pills for 12 months. Outcomes included: change in stress reactivity composite z-score (combining inflammatory, cortisol, and hemodynamic responses to a standardized psychological laboratory stressor); flow-mediated dilation (FMD) of the brachial artery (an index of vascular endothelial function); baroreflex sensitivity; and metabolic risk (presence of the metabolic syndrome or insulin resistance), all assessed at baseline and at months 6 and 12.

Results

Of 172 women enrolled, those assigned to TE + IMP tended to have higher resting baroreflex sensitivity than those assigned to placebo across the 6- and 12-month visits. Although treatment groups did not differ in terms of the other prespecified outcomes, a significant treatment-by-age interaction was found for FMD and stress reactivity such that an age-related decrease in FMD and increase in stress reactivity were seen among women assigned to placebo but not those assigned to TE + IMP. Women on TE + IMP also had lower resting diastolic blood pressure, lower levels of low-density lipoprotein cholesterol, and higher baroreflex sensitivity during stress testing.

Conclusions

TE + IMP tended to improve cardiac autonomic control and prevented age-related changes in stress reactivity and endothelial function among healthy perimenopausal and early postmenopausal women.

Cardiovascular disease (CVD), including arterial disease, stroke, and peripheral vascular disease, kills more women than men (1) and 10 times more women than breast cancer (2). The importance of midlife estradiol (E2) withdrawal as a contributing factor is supported by the rapid acceleration in the incidence of CVD after menopause (3–7) and by a large body of research linking other instances of E2 withdrawal (e.g., oophorectomy) with CVD and its precursors, both in humans (5, 8–15) and in animals (16–18). That midlife women experience a 2- to 3-fold increased risk of depression compared with younger women (19–22) also possibly contributes to the increased incidence of CVD, as suggested by research linking depression in this demographic with a 50% increase in cardiovascular mortality (5).

The Women’s Health Initiative (WHI) aimed to test the cardiovascular benefits of postmenopausal hormone therapy (conjugated equine estrogen [CEE] and medroxyprogesterone acetate [MPA]); however, its observation that hormone therapy actually increased the risk of breast cancer, venous thromboembolism, myocardial infarction, and stroke (23) dramatically limited the prescription and study of postmenopausal hormone therapy. Mounting evidence (24, 25) has since clarified that both the timing of hormone therapy initiation (26, 27) and the hormonal preparation used (28, 29) are important moderators of its cardiovascular effects, with transdermal estradiol and intermittent micronized progesterone (TE + IMP) initiated close to menopause onset being associated with the lowest risk (26, 28); however, its use remains controversial.

FDA-approved indications for systemic postmenopausal hormone therapy remain limited to vasomotor symptoms and the prevention of bone loss in women at high risk for osteoporosis (30). However, our recent reporting of the Perimenopausal Estrogen Replacement Therapy (PERT) Study (31), a 12-month double-blind, placebo-controlled randomized trial of TE (0.1 mg) and IMP in healthy perimenopausal and early postmenopausal women, suggests that hormone therapy may also have important mood benefits in this population, which is known to be at significantly increased risk of depressive symptoms (20, 21, 32–34) and clinical depression (19–22). Specifically, the PERT study observed that TE + IMP reduced the overall risk of clinically significant depressive symptoms from 32% to 17%, consistent with the few small trials finding E2 therapy to be an effective treatment for perimenopausal depression (35–37). The current manuscript reports on the PERT study outcomes of relevance to arterial disease risk to provide a more comprehensive picture of the overall health benefits and risks of TE + IMP in perimenopausal and early postmenopausal women. More specifically, it examines the effect of TE + IMP on 4 mechanisms involved in the early pathogenesis of atherosclerosis and arterial disease: (1) hemodynamic and neuroendocrine stress reactivity; (2) endothelial function, as indicated by flow-mediated dilation (FMD) of the brachial artery; (3) baroreceptor reflex sensitivity (BRS), a marker of cardiac autonomic control; and (4) metabolic risk, as indicated by the presence of the metabolic syndrome or insulin resistance. Furthermore, because age is the primary predictor of arterial disease for both men and women (38), secondary exploratory analyses examined participant age as a potential moderator of treatment effects.

Methods

Participants

From October 2010 to January 2015, 172 women aged 45 to 60 years, medically healthy and perimenopausal or early postmenopausal according to the Stages of Reproductive Aging Workshop (STRAW + 10) criteria (39) (within 2 years of the final menstrual period), were self-referred in response to advertisements posted throughout the community and on social media. Women without a uterus but with at least 1 ovary retained were included in the study if: (1) they were experiencing vasomotor symptoms and their baseline estradiol levels were above postmenopausal concentrations (> 40 pg/mL) or, (2) they were not experiencing vasomotor symptoms but had baseline estradiol > 40 pg/mL and baseline follicle-stimulating hormone levels > 14 pg/mL. However, these women did not take IMP and were not included in the reproductive stage moderation analyses. Women with both ovaries removed were included if the bilateral oophorectomy occurred in the past 24 months and they had been menstruating regularly prior to the procedure. These women were also not included in reproductive stage moderation analyses. All women underwent a pelvic exam with a study physician to screen for any signs or history of endometrial disorder or abnormal uterine or ovarian anatomy. An endometrial biopsy was performed to rule out endometrial cancer in cases of concerning bleeding patterns. The final participant completed the trial in February 2016. Data were collected at the University of North Carolina at Chapel Hill. The trial protocol was approved by the University of North Carolina’s Institutional Review Board. All participants provided informed, written consent prior to participating and received up to $1425 in compensation for participating in full compliance.

Trial design

The PERT study was designed to examine the effects of TE + IMP on depressive symptoms and arterial disease risk factors among healthy perimenopausal women and to investigate several mechanisms underlying estradiol’s effects. The current manuscript reports on the outcomes and mechanisms relevant to arterial disease. PERT used a randomized, double-blind, placebo-controlled design in which 172 women were enrolled and randomly assigned to treatment with either patches of 0.1 mg of 17β-estradiol or placebo patches (developed by 3M Pharmaceuticals, St. Paul, Minnesota) for 12 months. Oral micronized progesterone (200 mg/day for 12 days) was given every 2 months to women on active TE + IMP to protect the endometrium, and an identical schedule of placebo pills was implemented for women on placebo. However, if a participant experienced particularly bothersome mood, sedative, or other side effects with the progesterone, then it was given every 3 months, at the study physician’s discretion. Postrandomization study visits occurred at 1, 2, 4, 6, 8, 10, and 12 months; however most outcomes relevant to arterial disease were measured only at 6 and 12 months. To allow for the examination of the effects of TE without the confounder of progesterone, no visits occurred while women were taking progesterone. Further methodological details, including the original trial protocol, have been published elsewhere (32).

Measures

The primary outcome variable (stress reactivity composite z-score change) and the 3 secondary outcome variables (FMD, BRS, and metabolic risk) were assessed at baseline, 6 months postrandomization and 12 months postrandomization.

Stress reactivity composite z-score change.

Change in stress reactivity to the Trier Social Stress Test (TSST) (40), a psychosocial laboratory stressor, from prerandomization stress reactivity levels was the primary outcome of relevance to arterial disease assessed in the PERT study. At each laboratory session, the TSST followed instrumentation and intravenous set-up, a 20-minute recovery from venipuncture and a 10-minute baseline rest period. The TSST involved 4 components: (1) pre-task instructions (1 minute) during which participants were introduced to a panel of judges who would listen to their job talk; (2) speech preparation (5 minutes) during which time participants prepared their job talk while the selection committee remained present; (3) job speech (5 minutes) immediately following the preparation period, during which the panel asked the participant to deliver her job talk; and (4) serial subtraction task, which involved serially subtracting a 1-digit number from a 4-digit number (e.g., 7 from 2000) as fast and as accurately as possible for 5 minutes. The TSST was followed by a 60-minute recovery period. A different job talk and set of numbers was used at each visit to limit habituation, as suggested by Foley and Kirschbaum (41).

Blood pressure and total peripheral resistance were measured every 2 minutes during the baseline rest period (i.e., minutes 21, 23, 25, 27, and 29 following instrumentation) and were averaged to yield a mean baseline value. Levels assessed at minutes 0 and 3 of the job speech were averaged to yield a mean stress value, since the speech portion of the TSST is typically associated with the greatest cardiovascular response (42, 43). Serum cortisol and the inflammatory marker interleukin-6 (IL-6) were measured at the end of the baseline rest; stress levels were measured 10 minutes following the end of the serial subtraction task, thus capturing the cortisol response that typically peaks 20 to 25 minutes following the beginning of the TSST (42, 43). The SunTech exercise blood pressure monitor, Model 4240 (SunTech Medical Instruments, Inc., Raleigh, NC), provided blood pressure measurements, while a custom-designed impedance cardiograph (HIC-100 Bioimpedance Technology, Inc., Model 100, Chapel Hill, NC, USA) was used in conjunction with a tetrapolar band electrode configuration to noninvasively assess total peripheral resistance. To limit the family-wise error rate due to multiple comparisons, IL-6, cortisol, mean arterial pressure, and total peripheral resistance, indexed for body size, were used to form a stress reactivity composite z-score as follows: the difference between resting and stress levels of each measure was calculated, yielding a reactivity score for each measure; these reactivity scores were then each transformed into z-scores and averaged to create a single composite z-score.

Flow-mediated dilation (FMD) and nitroglycerin-mediated dilation of the brachial artery.

Ultrasound images of the right brachial artery were acquired proximal to the antecubital fossa by a registered sonographer using an HDI 5000 ultrasound machine (Philips Medical; Boston, MA) interfaced with a 12 mHz transducer. Arterial diameters were measured at end-diastole using customized wall-tracking software (Vascular Analysis Tools, Medical Imaging Applications, LLC). FMD, a measure of endothelial function, was assessed by determining the percent change in arterial diameter in response to reactive hyperemia 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 90 seconds following cuff deflation. After a 15-minute rest, a second baseline image was acquired, and sublingual nitroglycerin (0.4 mg) was administered to determine endothelium-independent vasodilation, and ultrasound images were acquired for the subsequent 5 minutes. FMD and nitroglycerine-mediated dilation were calculated as percentage changes in diameter from their respective baselines.

Baroreceptor reflex sensitivity (BRS).

Beat-by-beat systolic blood pressure (SBP) and pulse rate were collected using the Finometer noninvasive blood pressure monitor (Finometer Pro Model 1, Finapres Medical Systems, Nieuwkoop, Netherlands). Cross-spectral analysis was used to estimate BRS. For these analyses, beat-by-beat blood pressure and interbeat interval (derived as 60 000/heart rate) were edited for artifacts, linearly interpolated, and resampled at a frequency of 4 Hz, in order to generate an equally-spaced time series. A fast Fourier transformation was applied to the interpolated data after the detrending process and then a Hanning filtering window was applied. Power spectra were derived for each file using the Welch algorithm, which ensemble averages successive periodograms (44). BRS was estimated from the magnitude of the transfer function relating interbeat interval oscillations to SBP oscillations across the 0.07 to 0.1299 Hz band. Coherence between SBP and interbeat interval oscillations was required to be at least 0.5 in order to be accepted as estimates of baroreflex control.

BRS was measured during 4 separate 5-minute intervals: the final 5 minutes of the baseline rest period, the speech preparation period, during the speech, and at minute 30 of the recovery period. BRS during the 5-minute resting period was prespecified as the secondary outcome of interest in the original trial protocol; however, treatment effects on BRS, assessed across all 4 testing intervals, were also examined.

Metabolic Risk.

Participants were classified as having metabolic risk if they either: (1) met standard National Cholesterol Education Program (NCEP) criteria (45) for metabolic syndrome based on any 3 of the following risk factors: (a) blood pressure ≥130/ ≥85 mm Hg, (b) waist circumference > 88 cm, (c) triglycerides ≥ 150 mg/dL, (d) high-density lipoprotein (HDL) cholesterol < 50 mg/dL, or (e) fasting glucose ≥ 100 mg/dL; or (2) were insulin-resistant (a clear metabolic predictor of cardiovascular disease), as defined by a homeostatic model assessment of insulin resistance (HOMA-IR) score > 3.2 (46).

Resting blood pressure was assessed using a SunTech exercise blood pressure monitor, Model 4240 (SunTech Medical Instruments, Inc., Raleigh, NC), using a minimum of 2 resting measurements, acquired 5 minutes apart. Waist circumference was measured using the NHANES III protocol (47). Triglycerides and cholesterol blood levels were assessed following an overnight fast. HOMA-IR was calculated using the following formula: [fasting insulin (microIU/L) × fasting glucose (nmol/L)/22.5].

Other predetermined outcome variables.

Other variables described in the original trial protocol, but not identified as primary or secondary outcomes, included resting blood pressure, assessed in-clinic at 1, 2, 4, 6, 8, 10, and 12 months, and HDL and LDL cholesterol assessed at 6 and 12 months. Measures obtained at months 6 and 12 during stress testing but not included in the stress reactivity composite score included heart rate, plasma epinephrine and norepinephrine, plasma adrenocorticotropic hormone (ACTH), and cardiac output.

Vasomotor symptom (VMS) bother.

In addition to the above-mentioned outcomes, VMS bother was measured during the enrollment session and at each study visit using the Greene Climacteric Scale (GCS), a self-report form that asks participants to rate the extent to which they are currently bothered by 21 menopausal symptoms on a 4-point scale from ‘not at all’ to ‘extremely,’ Two of these items included “hot flushes” and “sweating at night,” which were used to respectively assess hot flashes and night sweats in the current study. The addition of these 2 items was used to create a score for overall VMS bother.

Statistical analysis

Power Analyses.

The current RCT was powered to detect an effect of TE + IMP on depressive symptoms, another primary outcome included in the study and reported on elsewhere (31). Thus, we carried out power analyses for the outcomes reported in the current manuscript as sensitivity analyses. For each continuous primary outcome (stress reactivity score, FMD, and BRS), the observed intraclass correlation in our sample was used to estimate the smallest detectible effect size (48). Using effect size conventions for Cohen’s f as follows: 0.10 = small; 0.25 = medium; 0.40 = large, the smallest detectible effect sizes for all 4 primary outcomes were conventionally small to medium, ranging from 0.19–0.20. For metabolic risk, sensitivity analyses were carried out using G*Power: setting the alpha level at 0.05, and power at 80%, the smallest detectible effect size was OR = 0.55, which translates to a 10% reduction in risk associated with TE + IMP, assuming a base rate of 26%. Thus, we conclude that the current study was adequately powered to detect clinically meaningful changes in the primary outcomes of interest.

Unless otherwise specified, an intention-to-treat analysis was performed. First, we tested the main effect of treatment (placebo versus TE + IMP) on the outcomes. Second, we tested the interaction between treatment and age.

All analyses were conducted using SAS 9.4 (SAS Institute; Cary, NC). For continuous outcomes, a repeated-measures analysis with PROC MIXED (for mixed models) was used with either 7 measures (months 1, 2, 4, 6, 8, 10, and 12) or 2 repeated measures (months 6 and 12), depending on the outcome. The pretreatment baseline value of the outcome variable was included as a covariate, except in the case of stress reactivity composite z-score change, since the outcome already reflected change from baseline. For analyses in which FMD was the outcome, resting arterial diameter was also included as a covariate. Models testing the interaction between treatment and age additionally included baseline menopausal status (early perimenopausal, late perimenopausal, or early postmenopausal, according to the STRAW+10 criteria [39]) as a covariate. A first-order autoregressive covariance structure was specified for within-person error. The Kenward-Roger method was used for computing degrees of freedom for tests of fixed effects. For significant treatment-by-age interaction effects, the continuous effect of age was examined separately in the TE + IMP and placebo groups. Since mixed models do not delete missing data listwise, all available data were used. Logistic regression was used to examine the effect of treatment (and potential moderation by age) on metabolic risk. Finally, since VMS have been shown to be related to altered autonomic function and increased CVD risk (49–51), sensitivity analyses were carried out, testing whether any observed significant effects remained significant when statistically adjusting for the degree of change in VMS bother experienced since the prerandomization visit.

Results

Participant characteristics

One hundred and seventy-two women entered the trial; the CONSORT flow diagram has been published elsewhere (31). As seen in Table 1, women randomized to TE + IMP and placebo did not significantly differ in terms of any baseline demographic or cardiovascular variables.

Table 1.

Participant Characteristics

	Placebo (n = 86)	TE + IMP (n = 86)
Demographic Information
Age, y (SD)	51.0 (3.2)	51.0 (3.0)
Race, No. (%)
Caucasian	60/86 (70)	70/86 (81)
African American	20/86 (23)	13/86 (15)
Other	6/86 (7)	3/86 (4)
Educationa, mean (SD)	7.0 (1.1)	6.8 (1.1)
Incomeb, mean (SD)	9.0 (3.1)	8.7 (3.3)
Reproductive Stage, No. (%)
Early Peri	18/86 (21)	18/86 (21)
Late Peri	48/86 (56)	51/86 (59)
Early Post	20/86 (23)	17/86 (20)
Plasma estradiol, pg/mL	90.1 (55.2)	104.7 (102.2)
Current smoker, No. (%)	6/86 (7)	4/86 (5)
Hot flash bother (SD; 0 = not at all; 3 = extremely)	1.1 (0.9)	1.0 (0.9)
Night sweat bother (SD; 0 = not at all; 3 = extremely)	1.1 (0.9)	0.9 (1.0)
Baseline Arterial Disease Risk Factors
Resting SBP (mmHg)	112.0 (11.9)	112.5 (11.3)
Resting DBP (mmHg)	70.7 (7.4)	70.1 (8.3)
BMI (kg/m2)	25.8 (3.7)	25.5 (3.6)
Waist circumference (cm)	88.4 (13.7)	87.2 (13.8)
LDL cholesterol (mg/dL)	115.4 (3.2)	119.9 (29.7)
HDL cholesterol (mg/dL)	70.0 (1.8)	67.4 (16.4)
Fasting insulin (mIU/L)	12.2 (5.3)	13.0 (6.0)
Fasting glucose (mg/dL)	87.4 (9.5)	89.1 (8.9)
HOMA-IR (mass units)	2.7 (1.4)	2.9 (1.5)
Elevated metabolic risk, No. (%)	26/86 (24)	23/86 (28)
Fasting triglycerides (mg/dL)	79.3 (37.3)	88.1 (47.9)
FMD (%)	7.1 (5.2)	7.0 (4.7)
NMD (%)	28.1 (9.9)	28.1 (9.2)
Baseline Measures Obtained during Stress Testing (averaged across all testing phases)
SBP (mmHg)	127.7 (20.7)	125.9 (19.6)
DBP (mmHg)	77.7 (10.5)	77.4 (10.7)
Heart rate (bpm)	74.7 (13.6)	74.7 (15.3)
Cardiac output (l/min)	7.0 (2.0)	6.8 (2.1)
Total peripheral resistance (mmHg*min*mL-1)	1140.2 (363.4)	1193.8 (372.1)
Cortisol (μg/dL)	7.6 (4.4)	7.6 (4.5)
ACTH (pg/mL)	65.3 (24.1)	68.8 (33.0)
Epinephrine (pg/mL)	54.2 (79.0)	61.8 (58.3)
Norepinephrine (pg/mL)	441.7 (171.4)	424.4 (147.7)
IL-6 (pg/mL)	1.4 (0.9)	1.3 (0.5)
Baroreceptor sensitivity (msec/mmHg)	5.2 (2.6)	5.2 (2.8)

Abbreviations: ACTH, adenocorticotropic hormone; DBP, diastolic blood pressure; FMD, flow-mediated dilation; HDL, high-density lipoprotein; HOMA-IR, homeostatic model assessment of insulin resistance; IL-6, interleukin-6; LDL, low-density lipoprotein; NMD, nitroglycerine-mediated dilatation; SBP, systolic blood pressure.

^a6 = some college (including completion of junior college); 7 = graduated from 4-year college.

^b8 = $50 000 to $79 999; 9 = $80 000 to $99 999.

Stress reactivity composite Z-score change

It should be noted that the effect of treatment on the main outcomes was not found to significantly differ by study visit (months 6 vs 12), as indicated by nonsignificant treatment-by-time interaction effects (P > 0.05); thus, the average of all outcome variables across all postrandomization study visits are reported throughout the manuscript.

The main effect of treatment on stress reactivity composite z-score change across visits 6 and 12 was not significant (mean [standard error (SE)] = 0.00 [0.06] vs −0.15 [0.06]; P = 0.11). However, a significant interaction between treatment and age was found (P for the interaction = 0.04; Fig. 1A) such that, in the placebo arm, each additional year was associated with a statistically significant composite z-score increase of 0.08 (P < 0.01) but that in the TE + IMP arm, each additional year was associated with a nonsignificant change of −0.02 (P = 0.44).

Figure 1.

Stress reactivity score change (A) and flow-mediated dilation (B), across postrandomization months 6 and 12, by treatment and participant age category: 45–49 (n = 53), 50–54 (n = 83), and 55–60 (n = 11). Caption: For both outcomes, a significant interaction between treatment and age was found (P < 0.05) such that age was associated with significantly lower FMD and greater stress reactivity among women on placebo (P < 0.01) but not among women on TE + IMP (P > 0.41). *P < 0.05

Flow-mediated dilation

Adjusting for pretreatment baseline levels as well as resting arterial diameter, the main effect of treatment on FMD or nitroglycerin-mediated dilatation across visits 6 and 12 was not statistically significant (Table 2). However, a significant interaction between treatment and age on FMD was found (P for interaction = 0.04; Fig. 1B) such that in the placebo arm, each additional year was associated with a statistically significant decrease in FMD of 0.43 (P < 0.01) but that in the TE + IMP arm, each additional year was associated with a nonsignificant increase of 0.13 (P = 0.41). The interaction between treatment and age on nitroglycerin-mediated dilatation was not significant (P for the interaction = 0.22), suggesting that the treatment-by-age interaction effect on FMD is not driven by an endothelium-independent vasodilator response.

Table 2.

Mean (SE) of Arterial, Hemodynamic and Neuroendocrine Stress Outcomes by Treatment Arm, Averaged Across Postrandomization Months 6 and 12

	Placebo (n = 69)	TE + IMP (n = 63)	P value
Non-Stress Test Measures
Resting SBP (mmHg)	110.6 (0.8)	108.5 (0.8)	0.05
Resting DBP (mmHg)	70.5 (0.6)	68.8 (0.6)	0.04
BMI (kg/m2)	25.6 (0.1)	25.7 (0.1)	.56
Waist circumference (cm)	89.2 (0.5)	89.3 (0.5)	0.85
LDL cholesterol (mg/dL)	119.0 (1.7)	107.9 (1.8)	<0.0001
HDL cholesterol (mg/dL)	68.2 (0.9)	68.1 (0.9)	0.95
Fasting insulin (mIU/L)	12.8 (0.5)	11.4 (0.5)	0.04
Fasting glucose (mg/dL)	89.8 (0.9)	87.7 (0.9)	0.11
HOMA-IR (mass units)	2.9 (0.1)	2.4 (0.1)	0.01
Insulin resistance (HOMA ≥ 3.2) %	13.6%	8.8%	0.23
Fasting triglycerides (mg/dL)	89.2 (3.5)	88.6 (3.6)	0.90
FMD (%)	6.4 (0.5)	7.1 (0.5)	0.29
NMD (%)	30.0 (0.9)	30.0 (0.9)	0.99
Measures Obtained during Stress Testing (averaged across all testing phases)
SBP (mmHg)	125.2 (0.8)	126.4 (0.9)	0.32a
DBP (mmHg)	77.1 (0.5)	76.4 (0.5)	0.29
Heart rate (bpm)	75.3 (0.6)	74.3 (0.7)	0.30
Cardiac output (l/min)	6.9 (0.1)	7.2 (0.1)	0.06
Total peripheral resistance (mmHg*min*mL-1)	1198.2 (21.9)	1102.3 (23.5)	0.003
Cortisol (μg/dL)	7.3 (0.3)	7.0 (0.3)	0.40
ACTH (pg/mL)	61.1 (1.3)	59.7 (1.3)	0.45
Epinephrine (pg/mL)	48.6 (3.7)	39.4 (3.6)	0.06
Norepinephrine (pg/mL)	408.6 (12.0)	393.3 (12.1)	0.37
IL-6 (pg/mL)	1.36 (0.04)	1.40 (0.04)	0.50
Baroreceptor reflex sensitivity (msec/mmHg)	5.2 (0.2)	5.7 (0.02)	0.01

Abbreviations: ACTH, adenocorticotropic hormone; DBP, diastolic blood pressure; FMD, flow-mediated dilation; HDL, high-density lipoprotein; HOMA-IR, homeostatic model assessment of insulin resistance; IL-6, interleukin-6; LDL, low-density lipoprotein; NMD, nitroglycerine-mediated dilatation; SBP, systolic blood pressure.

^aNote: As there were no significant treatment-by-phase interactions, only the P values examining the main effect of treatment, averaged across all phases, are provided.

Baroreceptor reflex sensitivity (BRS)

Adjusting for pretreatment baseline levels, resting BRS across visits 6 and 12 tended to be higher in the TE + IMP versus placebo group (mean [SE] = 5.1 [0.2] vs 5.7 [0.2], respectively; P = 0.05). This effect was statistically significant when BRS was measured across all stress testing phases (Table 2; Fig. 2A; P = 0.01). Treatment did not significantly interact with stress testing phase (F[3, 620] = 0.2; P = 0.87), nor did treatment interact with participant age (β[standard error of the mean (SEM)] = −0.09 [0.11]; P = 0.42) to predict resting BRS.

Figure 2.

Baroreflex sensitivity (A) and total peripheral resistance (B), at baseline and across postrandomization months 6 and 12, by treatment. Caption: Adjusting for prerandomization levels, a significant main effect of treatment was found for both outcomes (P < 0.05) such that women assigned to TE + IMP had significantly higher baroreflex sensitivity (P = 0.01) and lower total peripheral resistance (P < 0.01) across all stress testing phases compared with women on placebo.

Metabolic risk

Adjusting for baseline number of metabolic risk factors, the percentage of participants with metabolic risk (defined as the presence of metabolic syndrome or insulin resistance) during the 12-month study was not significantly different in the TE + IMP group versus the placebo group (42% vs 39%, respectively; P = 0.70). There was also no significant interaction between treatment and age (β(SEM) = −0.01(0.08); P = 0.93).

Other predetermined outcome variables

As seen in Table 2, women assigned to TE + IMP had an overall healthier cardiovascular profile based on several variables that were not considered main outcomes: lower resting DBP, fasting insulin, LDL cholesterol, and HOMA-IR (P < 0.05). TE + IMP was also associated with lower total peripheral resistance across all stress testing phases (P < 0.05; Fig. 2B) and tended to be associated with higher cardiac output and lower overall epinephrine. No significant treatment-by-age interaction effects on these outcomes were observed (P > 0.05).

Sensitivity analyses: change in VMS bother as a confounding variable

TE + IMP was found to significantly reduce VMS bother more than placebo (P < 0.001). Therefore, for all statistically significant effects reported above, a second analysis was conducted that included change in VMS bother from baseline as a covariate. In all cases, effects remained, with the exception of TE + IMP’s effect on resting DBP, which fell to a trend level (P = 0.09).

Discussion

The current study examined the effects of TE + IMP on arterial disease risk factors, compared with placebo, among healthy perimenopausal and early postmenopausal women ages 45 to 60 years. Women assigned to TE + IMP exhibited a healthier profile characterized by higher BRS, lower DBP, lower LDL cholesterol, lower fasting insulin and improved insulin sensitivity. Significant age-by-treatment effects further suggested that TE + IMP eliminated the relationship between advancing age and both decreasing FMD and increasing stress reactivity, which were evident in the placebo group.

These findings are consistent with accumulating research suggesting that while hormone therapy may be harmful in women who are older or who initiate therapy with a long latency following onset of menopause, beneficial or neutral effects on cardiovascular health are seen in women who initiate therapy close to the menopause onset (26, 27, 52–56). This “timing hypothesis,” emerging following a reanalysis of the WHI (55), and largely based on primate research by Clarkson and colleagues (57), has since been supported in multiple studies: a Cochrane review from 2015 (26) identified 6 studies examining the effects of hormone therapy for the prevention of cardiovascular disease in postmenopausal women who began hormone therapy less than 10 years after menopause. Overall, hormone therapy (4/6 studies used oral CEE and 2 used oral E2) reduced all-cause mortality (RR (95% CI) = 0.70 (0.52–0.95) and coronary heart disease (RR (95% CI) = 0.52 (0.29–0.96) in this subgroup, though it did increase venous thromboembolism risk (RR (95% CI) = 1.74 (1.11–2.73).

While the current study did not directly compare TE and CEE, the hormonal preparation used in the WHI study, its findings are consistent with mounting evidence that TE may have more positive cardiovascular effects than CEE (see (28) for review). In 1 randomized controlled trial (RCT) comparing the endothelial and hemodynamic effects of 6 months of oral CEE (0.625 mg/day) and TE (0.05 mg/day), plus daily MPA, among 82 healthy postmenopausal smokers, only the TE group exhibited improvements in FMD; those assigned to TE also exhibited a greater reduction in norepinephrine, blood pressure, and TPR in response to mental stress (58). A second RCT compared 12 months of TE (0.36 mg/day) or oral CEE (0.625 mg/day), plus cyclic MPA, on arterial stiffness and vascular inflammatory markers in 28 postmenopausal women (59). TE was found to decrease arterial stiffness and circulating levels of vascular inflammatory markers; oral CEE, on the other hand, did not change arterial stiffness and had mixed effects on inflammatory markers. These findings are consistent with previous studies that had found oral CEE to have no effect on arterial stiffness (60, 61) and TE to have beneficial effects (62, 63). Meta-analytic evidence also suggests that, unlike oral estrogen, TE is not associated with an increased risk of venous thromboembolism (64, 65).

Similarly, although the current study did not directly compare micronized progesterone with alternative formulations, our findings are consistent with mounting research suggesting that micronized progesterone may be safer than synthetic progestins, such as MPA (see (28) for review). Among women ages 50 to 59 in the WHI, CEE alone but not CEE + MPA was associated with fewer coronary events (55), suggesting that MPA may counter the beneficial cardiovascular effects of estrogens. However, mounting evidence suggests that the same may not be true of micronized progesterone: for example, the Kronos Early Estrogen Prevention Study (KEEPS) trial (66) observed beneficial effects of estrogen therapy (CEE or TE) on lipid profiles and insulin resistance despite being paired with IMP. The Postmenopausal Estrogen/Progestin Interventions (PEPI) trial (67) also found that postmenopausal CEE + IMP resulted in a more favorable lipid profile when compared with CEE + MPA. IMP may also be safer for the breast as research in both animals (68) and humans (69, 70) suggests that micronized progesterone does not increase proliferation of breast tissue as do synthetic progestogens like MPA.

Further research is needed to confirm the optimal progesterone regimen needed to minimize the risk of endometrial hyperplasia while also minimizing unnecessary exposure to progesterone-related side effects, potential reversal of TE’s beneficial cardiovascular and metabolic benefits, and increased breast cancer risk. The current study administered IMP every 2 months, unless bothersome side effects were present, in which case IMP administration every 3 months was permitted. The decision to allow such a regimen is based on both uncontrolled trials (71–73) and one RCT (74) suggesting that progestin given every 2 to 3 months has a similar endometrial safety profile as progestin given every month. However, it should be noted that the Scandinavian Long Cycle Study Group did observe an increased incidence of endometrial hyperplasia among women receiving progestin every 12 weeks versus 4 weeks (5.6% vs 1%) (75), suggesting that further research aimed at identifying the optimal hormonal regimen associated with the safest overall profile is warranted.

The current study’s results, combined with the above-described data suggesting that TE + IMP is safe in young perimenopausal and early postmenopausal women, are encouraging in light of our recent report that TE + IMP has clinically significant mood benefits in this population (31), which is at increased risk for major depression (19–22). Specifically, women assigned to placebo were 2.5 times more likely to develop clinically significant depressive symptoms when compared with women assigned to TE + IMP; women in the early menopause transition and women reporting multiple stressful life events were found to experience an even greater mood benefit from treatment. The findings reported here provide reassurance that TE + IMP may be used to prevent the development of clinically significant depressive symptoms without adverse cardiovascular effects.

The findings of the current study must be interpreted in light of some limitations. First, it was not powered to detect treatment effects on hard cardiovascular end-points. However, cardiovascular reactivity (76–80), diminished BRS (81), and endothelial dysfunction (82–87) have all been shown to predict cardiovascular morbidity and mortality in both healthy and diseased populations. The current study was also not powered to detect treatment effects on cancer diagnosis or surrogate markers of cancer risk; however, the combined findings from multiple trials suggest that the association between breast cancer and hormone therapy use is very small (less than 0.1% per year) (88). Furthermore, as mentioned above, micronized progesterone does not increase proliferation of breast tissue as do synthetic progestogens like MPA (69). Thus, even this small increase in breast cancer risk likely does not apply to TE + IMP. Finally, it should be emphasized that the current study’s findings may not translate to a diseased population. Given our healthy sample, it is unsurprising that most mean outcome values remained in the normal range for participants in both treatment arms. While some of the statistically significant effects outlined in Table 2 are quite small and may be of limited clinical significance (e.g., resting DBP), other effects, such as the effect of TE + IMP on LDL cholesterol, may be clinically important: indeed, there is evidence that, at least among diabetic individuals without dyslipidemia, a 10-mg/dL difference in LDL cholesterol is associated with a 12% change in risk for cardiovascular disease (89). It is also possible that some of the relatively small effects observed in this 12-month study may have become more robust over a number of years.

Conclusions

The current study found that TE + IMP administration may have some beneficial effects on stress reactivity, BRS, and endothelial function in healthy perimenopausal and early postmenopausal women. These findings suggest that, consistent with earlier observational studies, TE + IMP administered close to menopause onset may have arterial benefits for healthy women.

Glossary

Abbreviations

BRS

baroreceptor reflex sensitivity

CEE

conjugated equine estrogen

CVD

cardiovascular disease

DBP

diastolic blood pressure

E2

estradiol

FMD

flow-mediated dilation

HDL

high-density lipoprotein

HOMA-IR

homeostasis model assessment for insulin resistance

IL-6

interleukin 6

IMP

intermittent micronized progesterone

LDL

low-density lipoprotein

MPA

medroxyprogesterone acetate

PERT study

Perimenopausal Estrogen Replacement Therapy study

SBP

systolic blood pressure

SE

standard error

SEM

standard error of the mean

STRAW + 10

Stages of Reproductive Aging Workshop

TE

transdermal estradiol

TSST

Trier Social Stress Test

VMS

vasomotor symptoms

WHI

Women’s Health Initiative

Additional Information

Disclosure Summary: The authors have nothing to disclose.

Data Availability: Restrictions apply to the availability of data generated or analyzed during this study to preserve patient confidentiality or because they were used under license. The corresponding author will on request detail the restrictions and any conditions under which access to some data may be provided.