Determinants of attained estradiol levels in response to oral estradiol plus progesterone therapy

Abstract

Objectives:

Among postmenopausal women taking hormone therapy (HT), the estradiol (E2) dose and E2 levels were differentially associated with change in metabolic measures. We evaluated determinants of attained E2 levels in response to HT.

Methods:

Postmenopausal women from the REPLENISH trial tested four formulations of oral combined E2 and progesterone compared with placebo. Mixed-effects linear models assessed characteristics associated with E2 levels among women with ≥80% HT compliance, adjusted for E2 dose and baseline E2 level.

Results:

Among 1173 postmenopausal women with mean (standard deviation) age 55 (4.3) years and 5.2 (4.8)years since menopause, higher treated E2 levels were significantly related to younger age, more recent menopause, and current alcohol use, while lower E2 levels were related to current smoking. Both age and time since menopause were significantly inversely associated with E2 levels; time since menopause had a stronger association with E2 levels. In the final multivariable model, E2 levels were positively associated with current alcohol use, and inversely associated with time since menopause and current smoking.

Conclusion:

Adjusting for E2 dose and baseline E2 level, on-trial E2 levels were significantly associated with time since menopause, current smoking, and current alcohol use. Practitioners should consider these factors in individual women to achieve a desirable E2 level during HT.

Introduction

Among postmenopausal women taking hormone therapy (HT), the blood estradiol (E2) level has been shown to be significantly associated with HT treatment efficacy on vasomotor symptoms, bone mass density, and changes in metabolic measures¹. According to the estrogen threshold hypothesis, different clinical endpoints have different response thresholds¹. For instance, vasomotor symptoms may be relieved with E2<15 pg/ml; increases in bone mineral density, high-density lipoprotein cholesterol and apolipoprotein A require E2≥15 pg/ml; and reductions in total cholesterol, low-density lipoprotein, and apolipoprotein B require E2≥25 pg/ml². The effect of HT on atherosclerosis progression is also correlated with the E2 level. In the Early vs. Late Intervention Trial with Estradiol (ELITE), higher E2 levels were associated with decreased atherosclerosis progression in early postmenopause, and in contrast were associated with increased atherosclerosis progression in late postmenopause³. Our prior analysis from the REPLENISH trial demonstrated that E2 dose and E2 levels were differentially associated with change in metabolic measures⁴. Despite the four different E2 doses studied, we found only a moderate correlation between oral E2 dose and serum E2 levels⁴.

Oral E2 is absorbed into the gastrointestinal system and metabolized in the liver through the cytochrome P450 enzyme pathway before entering the systemic circulation^5,6. The half-life of E2 is 13–20 h and excretion occurs through urine and feces⁷. We hypothesized that demographic and clinical characteristics relating to liver and kidney metabolism are determinants of the E2 level in postmenopausal women taking oral E2 therapy.

Methods

REPLENISH trial

The REPLENISH trial was a phase 3 randomized, double-blinded, placebo-controlled, multicenter trial testing a new formulation of HT, TX-001HR (TherapeuticsMD, Inc., Boca Raton, FL, USA), a combined E2 and natural progesterone (P4) in a single gelatin capsule with similar bioavailability to individual oral E2 and P4⁸. The REPLENISH trial was designed to determine the efficacy of TX-001HR on reduction of vasomotor symptoms (VMS substudy) and endometrial safety compared to placebo⁹. The trial was registered with ClinicalTrials.gov as NCT01942668.

Eligible participants were healthy postmenopausal women aged 40–65 years with a uterus who were seeking treatment for vasomotor symptoms, with serum E2≤50 pg/ml, with body mass index (BMI)≤34 kg/m², using ≤2 antihypertensive drugs, with a negative screening mammogram, and with normal breast examination and endometrial biopsy. Exclusion criteria were contraindications to HT, heavy smokers (use≥15 cigarettes per day), history of endometrial hyperplasia or undiagnosed vaginal bleeding, history of cancer, or clinically significant physical or mental illness.

A total of 1845 eligible participants were randomly assigned to one of five intervention groups of daily E2/P4 (1/100 mg, 0.5/100 mg, 0.5/50 mg, 0.25/50 mg) or placebo. The primary results of the REPLENISH trial were reported elsewhere^10-13.

The analysis reported here included REPLENISH participants who were randomly assigned to TX-001HR with at least 80% compliance assessed by pill count and who had available serum E2 levels at baseline and at least one follow-up visit at 1, 3, 6, 9, and 12 months.

Serum E2 level measurement

Serum E2 levels were centrally measured at screening and at post-randomization months 1, 3, 6, 9, and 12 using a single gas chromatography–tandem mass spectrometry method that was validated for human serum in the range of 2–500 pg/ml; the inter-assay coefficient of variation was <8.5% (Ventiv Health Clinical Lab Inc., Princeton, NJ, USA).

Potential correlates of E2 level

Potential correlates at baseline included in the analysis were age, time since menopause, weight, BMI, creatinine, creatinine clearance (calculated by Cockcroft–Gault method¹⁴: [140 – age (years)×weight (kg)×0.85] / [72×serum creatinine (mg/dl)]), aspartate aminotransferase (AST), alanine aminotransferase (ALT), current smoking status (yes/no), current alcohol use (yes/no), and use of dyslipidemia, antihypertensive, anticonvulsant, diabetic, and antifungal medications.

These variables were considered possible correlates of attained E2 levels, as they are related to E2 metabolism and excretion pathways, in particular, factors that are involved with cytochrome 450 enzyme in the liver or involved with kidney function.

Statistical analysis

Characteristics of women included in the analysis were reported as means and standard deviations (SDs) for continuous variables and frequencies and as percentages for categorical variables. Associations between on-trial E2 levels and each potential correlate were first assessed separately with mixed-effects linear models with a random intercept for each participant, adjusted for E2 dose (1, 0.5, 0.25 mg), baseline E2 level, and months of follow-up as the time variable. Measures with p≤0.15 from the analysis were included in the multivariable model. A backward selection approach was used to drop least-significant variables from the model; variables with p<0.05 were retained in the final model. The collinearity of independent variables in the final model was evaluated to avoid highly correlated variables in the final model. To test whether the associations of identified correlates were equally evident across E2 dose, the effect of the E2 dose on the association between each significant variable was also evaluated. Mean E2 levels with 95% confidence intervals were estimated from the multivariable model according to combinations of current smoking, current use of alcohol, and different time since menopause (1, 5, and 10 years) stratified by E2 dose (1, 0.5, and 0.25 mg). All statistical analyses were conducted using SAS software version 9.4 (SAS Institute Inc., Cary, NC, USA).

Results

Baseline characteristics

This analysis included 5881 study visits completed among 1173 postmenopausal women (1684 participants randomized to HT, 511 participants excluded due to <80% compliance) with mean (SD) age 54.7 (4.3) years and 5.2 (4.8) years since menopause. Women in the study had mean BMI of 26.7 (4.1) kg/m². Mean liver enzymes and creatinine levels were within normal ranges. Most women were White (67.6%) and currently used alcohol (64.8%). Current smoking was reported among 20.2% of women. Approximately 20% of women reported use of blood pressure medication (19.7%) or lipid medication (19.0%). A very small proportion of women reported use of anticonvulsant (5.4%), antidiabetic, systemic, or topical antifungal medications (<1%) (Table 1). The mean (SD) baseline E2 level was 6.15 (6.23) pg/ml and the mean on-trial E2 level was 27.71 (31.34) pg/ml.

Table 1.

Baseline characteristics of study participants.

Variable	Mean ± standard deviation or frequency (%)
Age (years)	54.8 ± 4.3
Time since menopause (years)	5.2 ± 4.8
Weight (kg)	72.5 ± 12.4
Body mass index (kg/m2)	26.7 ± 4.1
Aspartate aminotransferase (U/l)	18.1 ± 7.0
Alanine aminotransferase (U/l)	15.8 ± 8.7
Creatinine (mg/dl)	0.78 ±0.14
Creatinine clearance (ml/min)	95.6 ± 23.1
Race
White	794 (67.6%)
Black/African American	358 (30.5%)
Other	10 (0.9%)
Asian	7 (0.6%)
American Indian/Alaska Native	4 (0.3%)
Unknown	1 (0.1%)
Current smoking	238 (20.2%)
Current alcohol use	763 (64.8%)
Use of blood pressure medication	232 (19.7%)
Use of lipid medication	224 (19.0%)
Use of anticonvulsant medication	63 (5.3%)
Use of antidiabetic medication	8 (0.7%)
Use of systemic antifungal medication	2 (0.2%)
Use of topical antifungal medication	9 (0.8%)
Estradiol dose
0.25 mg	283 (24.0%)
0.5 mg	613 (52.1%)
1mg	281 (23.9%)

Associations between on-trial E2 levels and each baseline measure

On-trial E2 levels were positively correlated with AST, ALT, creatinine clearance, current alcohol use, use of blood pressure medication, and use of antidiabetic medication. Age, time since menopause, weight, BMI, creatinine, current smoking, and use of lipid, anticonvulsant, and antifungal medications showed negative correlations with on-trial E2 levels (Table 2).

Table 2.

Associations between on-trial serum estradiol level with each demographic and clinical characteristic.

Variable	β	SE	p-Value
Age (years)	−0.38	0.14	0.01
Time since menopause (years)	−0.55	0.12	<0.0001
Weight (kg)	−0.004	0.05	0.94
Body mass index (kg/m2)	−0.12	0.15	0.42
Aspartate aminotransferase (U/l)	0.10	0.06	0.11
Alanine aminotransferase (U/l)	0.06	0.05	0.29
Creatinine (mg/dl)	−5.99	3.83	0.12
Creatinine clearance (ml/min)	0.02	0.03	0.42
Estradiol dose (mg)
0.25 mg	Reference		<0.001
0.5 mg	8.83	1.46
1mg	28.05	1.71
Race
White	Reference		0.01
Black/African American	−4.87	1.30
Other	−2.22	6.50
Asian	−2.32	7.68
American Indian/Alaska Native	7.08	10.14
Unknown	−3.43	20.25
Current smoking	−4.13	1.47	0.005
Current alcohol use	4.27	1.24	0.0006
Use of blood pressure medication	2.75	1.49	0.07
Use of lipid medication	−2.50	1.51	0.10
Use of anticonvulsant medication	−0.48	2.63	0.86
Use of antidiabetic medication	4.88	7.21	0.50
Use of systemic antifungal medication	−9.02	14.37	0.53
Use of topical antifungal medication	−5.81	6.80	0.39

β estimates and p-values are from mixed-effects linear model, adjusted for estradiol dose and baseline estradiol level. SE, standard error.

Higher E2 levels were significantly related to higher E2 dose, younger age, more recent menopause, current alcohol use, and self-reported race/ethnicity as American Indian or Alaska Native (compared to White). Lower E2 levels were related to current smoking and self-reported African-American race (compared to White; p<0.05) (Table 2).

Independent variables selected for inclusion in the multivariable model (p<0.15) were age, time since menopause, AST, creatinine, race, current smoking, current alcohol use, and use of antihypertensive medication and dyslipidemia medication.

Correlation among variables included in multivariable modeling

Among predictors to be included in the multivariate model, age was significantly and highly correlated with time since menopause (r=0.54, p<0.0001). The correlations between other variables in the model were small (all correlations ≤0.16).

Multivariable analysis

In the multivariable model, both age (β [SE] = −0.43 [0.14] per month, p=0.002) and time since menopause (β [SE] = −0.52 [0.12] per month, p<0.0001) were significantly inversely associated with E2 levels. Including both variables in the model, age became non-significant (p=0.36), while time since menopause remained significant (β [SE] = −0.44 [0.15] per month, p=0.003) and had a stronger association with E2 levels than age (β [SE] = −0.16 [0.17] per month, p=0.36) (Table 3).

Table 3.

Multivariable models of association of on-trial serum estradiol level with significant correlated measure.

	Model with age			Model with time since menopause			Model with age and time since menopause
Variable	β	SE	p-Value	β	SE	p-Value	β	SE	p-Value
Age (years)	−0.43	0.14	0.002				−0.16	0.17	0.36
Time since menopause (years)				−0.52	0.12	<0.0001	−0.44	0.15	0.003
Current smoking	−4.83	1.48	0.001	−4.05	1.46	0.01	−4.31	1.49	0.004
Current alcohol use	4.10	1.23	0.001	3.92	1.23	0.002	3.91	1.23	0.002
Estradiol dose 0.5 mg	8.76	1.45	<0.0001	8.81	1.44	<0.0001	8.84	1.44	<0.0001
Estradiol dose 1mg	28.04	1.69	<0.0001	27.99	1.69	<0.0001	28.02	1.69	<0.0001
Baseline estradiol level (pg/ml)	0.46	0.10	<0.0001	0.46	0.09	<0.0001	0.45	0.10	<0.0001

β estimates and p-values are from mixed-effects linear model, adjusted for estradiol dose (dose = 0.25 mg as reference) and baseline estradiol level. SE, standard error.

As age and time since menopause were correlated, we developed the multivariable models with age and time since menopause modeled separately, along with other predictors. In each model, both older age (p=0.002) and longer time since menopause (p<0.0001) were associated with lower E2 levels. In all three models, current smoking was significantly associated with lower on-trial E2 level, while current alcohol use was significantly associated with higher on-trial E2 levels. AST, creatinine, race, and use of antihypertensive and dyslipidemia medications were not statistically significantly associated with E2 levels in multivariable modeling.

In the final multivariable model, time since menopause (p<0.0001), current alcohol use (p=0.002), and current smoking (p=0.002) significantly correlated with the on-trial E2 level when adjusted for E2 dose. On-trial E2 levels were positively associated with more recent time since menopause, current alcohol use, higher E2 dose, and higher baseline E2 levels but negatively associated with current smoking. E2 dose did not modify the association of time since menopause, current smoking, or current alcohol use with on-trial E2 level (all interaction p>0.05).

Predicted achieved E2 levels

The model-predicted mean E2 levels were higher with the 1mg E2 dose, followed by the 0.5mg and 0.25mg E2 dose, respectively. Depending on baseline smoking, alcohol, and time since menopause, predicted mean E2 levels were in the range 33.43–46.03pg/ml for a 1mg E2 dose, 14.26–22.93 pg/ml for a 0.5mg E2 dose, and 5.44–18.04 pg/ml for a 0.25mg E2 dose. Estimated E2 levels decrease with longer time since menopause. The predicted E2 levels were highest among postmenopausal women who currently used alcohol but did not smoke and were lowest among women who currently smoked but did not use alcohol (Figure 1).

Figure 1.

Estimated serum estradiol (E2) levels with E2 dose, current smoking, and alcohol use for time since menopause.

Discussion

Among postmenopausal women taking oral combined E2 and P4 with at least 80% compliance, E2 levels were positively associated with more recent time since menopause and current alcohol use, and inversely associated with current smoking when adjusting for baseline E2 level and E2 dose.

Both age and time since menopause were significantly associated with E2 levels. Postmenopausal women who were younger or had shorter time since menopause had higher E2 levels when taking HT compared with women who were older or had longer time since menopause. This finding was consistent with the correlates of baseline (untreated) E2 in the REPLENISH data that also showed lower E2 levels with older age¹⁵. Time since menopause had a stronger association with E2 levels than age at the time of HT initiation. This finding was consistent with a prior analysis from the ELITE that showed a significant difference in E2 levels by time since menopause among women taking oral E2 (p=0.02)³. In the ELITE, mean E2 levels were significantly higher in early postmenopause (<6 years since menopause) than in late postmenopause (≥10 years since menopause) despite similar compliance to daily oral E2.

The positive association of alcohol use and E2 level in these REPLENISH data is consistent with prior evidence that alcohol induces a rise in E2 levels in postmenopausal women. Alcohol consumption acutely increases E2 levels among postmenopausal women taking HT by 300% higher than the targeted level¹⁶. The association of alcohol use and E2 level has also been shown to be stronger with increasing dose of alcohol^17,18; dosage of alcohol use was not collected in the REPLENISH trial.

Proposed mechanisms of an alcohol effect on E2 level include an increased rate of aromatization of testosterone or a decreased rate of oxidation of E2 to estrone (E1)^19,20. Alcohol has been proposed to promote adrenal gland cell signaling for dehydroepiandrosterone sulfate production, which is then converted to E2^21,22. Alcohol is also associated with an increased nicotinamide-adenine dinucleotide NADH to NAD ratio in the liver²³, which leads to decreased catabolism of steroid hormones through oxidation and reduction of E2 conversion to E1¹⁶.

In a randomized cross-over study of oral and transdermal E2 therapy, E2 levels were 40–70% lower among smokers compared with non-smokers among postmenopausal women taking oral E2 therapy²⁴. Smoking can also reduce or completely cancel the clinical efficacy of oral E2 therapy, including alleviation of hot flushes and urogenital symptoms, and beneficial effects on lipid metabolism, osteoporosis, and cardiovascular disease^25,26.

A pharmacokinetics study reported enhanced hepatic metabolism of oral E2 with smoking²⁷; this supports the significantly reduced E2 levels among postmenopausal women in the REPLENISH trial who took HT and smoked. Smoking upregulates the 2-hydroxylation pathway of E2 metabolism and leads to decreased bioavailability of E2²⁸. Additional possible mechanisms of smoking-related reduction in E2 levels include smoking-related decreased activity of aromatase in granulosa cell and fatty tissue, reduction in steroid production from cholesterol through smoking-related inhibition of the C-20,22 desmolase enzyme, and inhibition of the cytochrome 450 enzyme system activity, alteration of sex-hormone binding globulin (SHBG) capacity, and adrenal function and renal clearance²⁶.

Several studies reported that E2 levels were positively associated with BMI in postmenopausal women taking HT^29-32. The mechanism was explained by increased aromatization of androstenedione to E1 and E2 with higher BMI^33,34. In the REPLENISH data, BMI was significantly positively associated with untreated baseline E2 level¹⁵. However, the current analysis did not find an association of BMI with the on-trial E2 level while taking oral E2 plus P4. These REPLENISH findings are similar to the results of a subgroup analysis from the Women’s Health Initiative trial showing that higher BMI was associated with increased baseline E2 but not with treated levels of E2 after 1 year of HT³⁵. A possible explanation would be that BMI is associated with endogenous E2 level at baseline, but is not associated with the metabolism of exogenous E2 level when taking HT in this group of population with BMI < 34 kg/m².

The strength of this study is the randomized clinical trial data providing prospective repeated measurements of E2 levels and accounting for the different doses (0.25, 0.5, and 1mg) of oral E2. However, the analysis was limited to total serum E2 level and did not account for other possible related levels of free E2, E1, and SHBG. We did not have data on levels of smoking and alcohol use. In addition, data on possible correlates of E2 were limited to baseline assessments; measurement of these variables at the time of the post-randomization E2 assessments could provide more informative data related to correlates of attained E2 levels.

The clinical implication of this study emphasizes the need for evaluation of factors related to the E2 level among postmenopausal women taking HT since these attained levels relate to the possible treatment efficacy of HT, in particular the change in metabolic measures⁴. Postmenopausal women should avoid smoking and alcohol use with the goal of obtaining the lowest effective dose of HT. Health-care providers prescribing HT need to consider time since menopause, current smoking, and current alcohol use and may adjust the treatment dose for each individual woman to maximize the treatment efficacy and minimize side effects.