Effects of Continuous Estrogen and Estrogen-Progestin Replacement Regimens on Cardiovascular Risk Markers in Postmenopausal Women FREE

THE BENEFICIAL effects of estrogen use in postmenopausal women for management of estrogen deficiency symptoms and prevention of bone loss have been widely documented.^1- 2 The addition of a progestin, especially in a continuous manner, has been shown to reduce the risk of endometrial cancer associated with unopposed estrogen use in women with an intact uterus.^3- 5

As cardiovascular disease is the most common cause of death in postmenopausal women, an important aspect to consider in the decision to use long-term postmenopausal estrogen replacement therapy (ERT) or estrogen-progestin replacement therapy (HRT) is its potential role in the prevention of coronary heart disease (CHD). A large number of epidemiological studies, including several meta-analyses, have shown a strong and consistent relationship between ERT or HRT use and reduced primary and secondary CHD risk.^6- 9 Nevertheless, limited data are available from randomized clinical trials assessing the influence of these therapies on CHD events.

In the first randomized, secondary prevention trial with HRT, the Heart and Estrogen-Progestin Replacement Study, no overall differences were found in the incidence of nonfatal myocardial infarction or CHD between groups treated with HRT (conjugated equine estrogens plus medroxyprogesterone acetate) and placebo.¹⁰ More CHD events were observed in the HRT group than with placebo in year 1 of treatment and fewer events in years 4 and 5 of treatment.¹⁰ Given the large body of epidemiological and animal data suggesting that ERT provides cardiovascular protection, the results of this trial highlight the need to identify the effects of progestins on cardiovascular health.

Previous clinical trials have shown that HRT affects several cardiovascular risk factors, such as lipoproteins, hemostatic variables, carbohydrate metabolism, and endothelial function.^5,11- 12 Various HRT products have differential effects on the metabolic cardiovascular risk profile, particularly with regard to the progestin used, underscoring the need for clinical trials to evaluate and compare HRT options. Norethindrone acetate is a 19-nortestosterone–derived progestin, with potent effects on minimizing estrogen-induced endometrial stimulation, even at very low doses.⁴ The available data from animal models suggest that norethindrone does not attenuate the antiatherosclerotic effects of estrogens.^3,13 Most of the information from humans on the metabolic effects of norethindrone has been obtained with high doses of 1 to 10 mg¹⁴; thus, there is a need to further assess the impact of low doses of norethindrone on cardiovascular measures. The present study was designed to evaluate the influence of 24 weeks of continuous low doses of norethindrone acetate, 0.25 and 0.5 mg, given with 1 mg of 17-β estradiol, compared with placebo and unopposed 17-β estradiol, 1 mg, on a wide range of risk markers for CHD in postmenopausal women.

The subjects for this trial were apparently healthy postmenopausal women, 70 years of age or less, with onset of natural or surgical menopause at least 12 months before enrollment. Postmenopausal status was confirmed by the presence of a plasma estradiol concentration of 73 pmol/L (20 pg/mL) or less.

Exclusion criteria included use of hormone replacement or lipid-altering agents within 8 weeks of the baseline (week −4) visit, body mass index (calculated as weight in kilograms divided by the square of height in meters%) greater than 31.5 kg/m², current use of more than 20 cigarettes per day, alcohol abuse (>14 alcoholic drinks per week), and other substance abuse. Women with uncontrolled hypertension (systolic blood pressure >160 mm Hg or diastolic blood pressure >95 mm Hg) or elevated triglyceride levels (>4.0 mmol/L [>350 mg/dL] at 2 consecutive visits) were excluded. Other medical conditions excluding participation were history of stroke, pancreatitis, gallbladder disease, thrombophlebitis, or thromboembolic disorders; myocardial infarction within 6 months; abnormal genital bleeding of unknown etiology; endometrial hyperplasia; an abnormal mammogram suggestive of malignant neoplasm; the presence of hepatic enzyme levels more than twice the upper limit of normal; diabetes mellitus or other endocrine disease (except hypothyroidism adequately treated with a stable dose of thyroid replacement); and notable psychiatric disorders. Women using β-adrenergic blockers, high doses of thiazide diuretics (>25 mg of hydrochlorothiazide per day or its equivalent), erythromycin, immunosuppressants, systemic corticosteroids, or anticoagulants, or who had participated in a clinical trial of another investigational agent within 30 days, were also not eligible to participate.

This randomized, double-blind, placebo-controlled trial with 4 parallel treatment arms was conducted at a single clinical research center (including a satellite location) in the Chicago, Ill, area. An institutional review board (Schulman Associates, Cincinnati, Ohio) approved the study protocol, and all subjects provided written informed consent. Subjects were recruited through advertisements and community outreach, were reimbursed for expenses, and received a monetary stipend for their participation.

Clinic visits were scheduled at 9 time points during the trial: weeks −4, −2, 0 (randomization), 3, 6, 12, 18, 23, and 24. At week 0, all eligible subjects were randomly assigned to 1 of 4 treatment groups: (1) placebo; (2) unopposed 17-β estradiol, 1 mg; (3) 17-β estradiol, 1 mg, combined with norethindrone acetate, 0.25 mg; or (4) 17-β estradiol, 1 mg, combined with norethindrone acetate, 0.5 mg (Activella; Novo Nordisk Pharmaceuticals, Inc, Princeton, NJ). The study medication and placebo were formulated in identical-appearing oral tablets and provided in 28-day, dial blister packs. Subjects were instructed to take 1 tablet with water each day at bedtime. Compliance was assessed by pill counts and subject interviews performed at each postrandomization clinic visit. A subject was considered noncompliant if she missed study medication for 6 or more days within any 28-day period.

All biochemical assays were completed at a central facility (Quest Nichols Institute, San Juan Capistrano, Calif). This central laboratory facility participates in the Centers for Disease Control and Prevention–National Heart, Lung, and Blood Institute lipid measurement standardization program.

Fasting blood samples were collected at weeks −2, 0, 12, 23, and 24 for assessment of serum levels of total cholesterol, high-density lipoprotein (HDL) cholesterol, triglycerides, and calculated low-density lipoprotein (LDL%) cholesterol. Additionally, lipoprotein(a) (Lp[a]) and apolipoproteins A-I and B (Apo A-I and Apo B) were measured at weeks 0, 12, and 23. Plasma cholesterol and triglyceride concentrations were determined with an analyzer (Hitachi 914; Boehringer Mannheim, Indianapolis, Ind) that uses enzymatic methods. The HDL cholesterol was quantified after precipitation of lower-density lipoproteins with phosphotungstate and magnesium. The LDL cholesterol level, in milligrams per deciliter, was calculated by means of the following equation: LDL cholesterol = total cholesterol-HDL cholesterol-triglycerides/5.¹⁵ This equation loses accuracy when the plasma triglyceride level exceeds 4.5 mmol/L (400 mg/dL). Accordingly, no LDL cholesterol value was calculated in cases where triglycerides were above this level. The Lp(a) concentrations were determined by means of an immunoprecipitation analysis with interpolation of sample and control concentrations to a calibration curve using 5 standards. Levels of Apo A-I and B were determined on an analyzer (Behring Nephelometer Analyzer II; Quest Nichols Institute) that assesses the turbidity resulting from specimen protein reaction with specific antisera to human Apo A-I and B, respectively.

Blood samples collected at weeks −2, 0, 12, and 23 were used to assess circulating levels of fibrinogen, plasminogen activator inhibitor 1 (PAI-1) antigen concentration, factor VII activity, and antithrombin III activity. Fibrinogen concentration was assessed with an automatic coagulation analyzer (model 900C; Medical Laboratory Automation, Pleasantville, NY) by means of a photo-optical clotting method. The PAI-1 concentrations were determined by color analysis of the reaction between a peroxidase enzyme substrate and the PAI-1 antigen-conjugate antibody in microtest wells (Quest Nichols Institute). Factor VII activity was determined with a clotting assay using photometric clot detection (MLA 1600; Medical Laboratory Automation). Antithrombin III activity was determined by spectrophotometry by means of a chromogenic substrate.

Blood samples collected at weeks −2, 0, 12, 23, and 24 were used for assessments of fasting glucose, insulin, and C-peptide concentrations. Hemoglobin A_1c (HbA_1c) was also measured in samples collected at weeks 0, 12, and 23. The serum glucose concentration was measured with an analyzer (Hitachi 914; Boehringer Mannheim) by the glucose oxidase method. The HbA_1c level was measured with an analyzer (VARIANT; Bio-Rad Laboratories, Hercules, Calif) by ion-exchange high-performance liquid chromatography. Plasma insulin concentration was assessed by radioimmunoassay (Linco Scientific, St Charles, Mo). Plasma C-peptide level was measured by a radioimmunoassay using a guinea pig anti–C-peptide antiserum (Quest Nichols Institute).

Estradiol concentration was measured in blood samples collected at weeks −4, 12, and 23. A competitive radioimmunoassay was used for the analysis of estradiol levels (Quest Nichols Institute). At weeks 0, 12, and 23, waist circumference (minimum circumference between the lower rib and iliac crest) was measured in duplicate or triplicate if the difference between the first 2 values was greater than 0.5 cm. The 2 closest values were recorded. At weeks 0, 12, and 23, subjects also completed 7-day Physical Activity Recall¹⁶ and Eating Pattern Assessment Tool¹⁷ questionnaires for evaluation of habitual physical activity and dietary patterns, respectively. On each study day, all adverse events, either observed by the investigator or spontaneously reported by the subject, were recorded and evaluated.

Statistical analyses were conducted by means of SAS (version 6.09) or Statview (version 5.0) statistical analysis packages (SAS Institute Inc, Cary, NC). Responses of cardiovascular risk markers to treatment were calculated for all subjects with evaluable data, defined as those who did not miss more than 5 doses of study medication during any 28-day cycle. Six subjects were excluded from the evaluable subset because of noncompliance. The change from baseline in LDL cholesterol concentration was defined in the study protocol as the primary outcome variable. Separate intent-to-treat analyses were also completed that included all subjects who took at least 1 dose of study medication and completed at least 1 postrandomization assessment. For these, the last observation available was carried forward to impute missing values. Response data from the evaluable and intent-to-treat samples did not differ materially; therefore, results presented are from the evaluable subset, which was defined, a priori, as the primary analysis.

The average of 2 measurements was used to define baseline (weeks −2 and 0) and end-of-treatment (weeks 23 and 24) values for total cholesterol, LDL cholesterol, HDL cholesterol, triglycerides, insulin, glucose, and C-peptide. Two samples were averaged at baseline (weeks −2 and 0) for hemostatic variables (fibrinogen, factor VII activity, PAI-1 antigen, and antithrombin III activity) and responses were calculated by subtracting the baseline value from the single value obtained at week 23. Changes from baseline to end of treatment for HbA_1c, Apo A-I, Apo B, and Lp(a) were calculated from measurements taken at weeks 0 and 23.

Comparability of treatment groups at baseline was assessed by analysis of variance for continuous variables and Pearson χ² tests for nominal variables. Analysis of variance models were used to evaluate treatment effects, with changes from baseline to the end of treatment used as dependent variables. Responses were not normally distributed for some variables (triglycerides, Lp[a], and PAI-1 antigen). In these cases, rank transformations were used before analysis of variance was performed. Results tables present data from these variables as median (minimum, maximum). All other continuous response variables are presented as mean (SD). Linear contrasts were used for pairwise comparisons between treatment groups when the F ratio for the corresponding analysis of variance model was at or above the critical value for the 5% level of statistical significance.

Of the 498 women screened, 270 were found to be eligible and were randomly assigned to 1 of the 4 treatment groups. The disposition of all subjects is shown in Figure 1.

Table 1 shows baseline characteristics of the evaluable sample (n = 264) according to treatment group assignment. No significant differences between treatment arms were noted with regard to race, hysterectomy or smoking status, alcohol intake, body mass index, waist circumference, systolic blood pressure, or diastolic blood pressure. The ages ranged from 42 to 70 years, with no significant differences between groups. Moreover, no significant differences between groups were observed at baseline for diet or physical activity habits as indicated by scores on the Eating Pattern Assessment Tool or the 7-day Physical Activity Recall, respectively. Scores on these questionnaires did not change significantly in any group during the trial (data not shown).

Table 2 presents data on lipid, lipoprotein, and apolipoprotein values for each group at baseline, week 12, and end of treatment and changes from baseline to the end of treatment. In addition, median percentage changes from baseline to the end of treatment are shown in Figure 2. No significant differences between groups were observed at baseline for any aspect of the serum lipid profile.

Mean total cholesterol concentration did not change significantly from baseline to end of treatment in subjects assigned to placebo (−0.1%). The total cholesterol response associated with unopposed 17-β estradiol (−2.3%) was not significantly different from the response in the placebo group. The groups treated with 17-β estradiol plus 0.25 and 0.5 mg of norethindrone acetate showed changes in total cholesterol level of −5.6% and −9.7%, respectively. Responses in the combined HRT regimens were significantly different from placebo (P<.01) and from one another (P<.05). Only the response in the 17-β estradiol–0.5-mg norethindrone acetate group reached significance compared with unopposed 17-β estradiol (P<.001).

The LDL cholesterol level showed a small decline from baseline in the placebo group (−2.4%). Significantly larger responses were observed in the 3 active treatment groups: −10.1% (unopposed 17-β estradiol; P = .001), −10.5% (17-β estradiol with 0.25 mg of norethindrone acetate; P = .004), and −13.5% (17-β estradiol with 0.5 mg of norethindrone acetate; P = .001), which did not differ from one another.

The mean HDL cholesterol concentration showed an unexpected 8.9% increase from baseline in the placebo group. Examination of the data indicated that this increase could not be attributed to changes in physical activity, Eating Pattern Assessment Tool scores, alcohol consumption, cigarette use, plasma estradiol concentrations, or concomitant medication use. Laboratory records also offered no explanation for this finding. Fasting lipid profiles were reanalyzed in batch from archived samples for each study participant. Results from the reanalysis did not differ materially from those of the original measurements, including the increase in HDL cholesterol concentration among subjects in the placebo arm. The HDL cholesterol level increased by 13.6% in the group that received unopposed 17-β estradiol, which did not differ significantly from placebo. Smaller mean increases were observed with 17-β estradiol plus 0.25 mg of norethindrone acetate (7.1%) and 17-β estradiol plus 0.5 mg of norethindrone acetate (3.4%). Only the HDL cholesterol response in the 0.5-mg group was significantly different from placebo (P<.01). Changes from baseline in the unopposed 17-β estradiol group differed significantly (P<.01) from those of both combined HRT arms, but the responses in the combined HRT arms were not significantly different from one another.

The triglyceride concentration remained similar to baseline level among subjects taking placebo (median change of −1.0%). The group that received unopposed 17-β estradiol showed a median increase of 9.4% in triglyceride concentration, which was significantly different from placebo (P<.05). Small changes in the median triglyceride concentration were observed with 17-β estradiol plus 0.25 mg of norethindrone acetate (4.9%%) and 17-β estradiol plus 0.5 mg of norethindrone acetate (−6.1%), which were not statistically different from one another or the response in the placebo group. However, the response of the group that received 17-β estradiol plus 0.5 mg of norethindrone acetate did differ significantly from that of the unopposed 17-β estradiol group (P<.01).

Median Lp(a) concentration increased in the placebo group (13.4%) and remained unchanged in the 3 active treatment arms. Despite the modest change in median value, the differences in response between placebo and the 17-β estradiol plus 0.25 mg (P<.05) and 17-β estradiol plus 0.5 mg (P<.05) groups reached statistical significance.

Small reductions from baseline in mean Apo B concentration were observed in the placebo (−1.6%) and unopposed 17-β estradiol (−1.1%%) groups. Changes from baseline were −5.4% and −7.7% in the 0.25-mg and 0.5-mg arms, respectively. These changes did not differ from one another, but that of the 0.5-mg group differed significantly from both placebo (P<.01) and unopposed 17-β estradiol group responses (P<.01).

Mean Apo A-I concentration declined slightly in the placebo group (−1.0%%) and increased by 8.9% in the unopposed 17-β estradiol group (P<.01 vs placebo). The change from baseline in Apo A-I level within the group that received 17-β estradiol plus 0.25 mg of norethindrone acetate (−1.1%) was similar to that in the placebo arm. 17-β Estradiol plus 0.5 mg of norethindrone acetate reduced Apo A-I level by −4.3% (P<.01 vs placebo). Responses in the 2 combined HRT groups did not differ from one another, but they did differ significantly from that of the unopposed 17-β estradiol group (P<.01).

Data from baseline, week 12, and end of study and changes from baseline to the end of treatment in hemostatic variables are shown in Table 3. Median percentage changes from baseline to the end of treatment are shown in Figure 3.

No significant differences in any of the hemostatic variables measured were observed between groups at baseline. A significant increase in mean fibrinogen concentration (12.4%) occurred in the placebo group during the 24-week treatment period. Smaller increases were observed in the 3 active treatment arms. The unopposed 17-β estradiol group displayed a mean change of 5.4%, which was significantly less than that of the placebo group (P<.05). The mean increase in the 17-β estradiol–0.25-mg norethindrone acetate group was 5.1%, which was near significance in comparison with the response in the placebo group (P<.07%) but did not differ from the unopposed 17-β estradiol or 17-β estradiol–0.5-mg norethindrone acetate groups. The change from baseline within the 17-β estradiol–0.5-mg norethindrone acetate arm (9.1%) did not differ significantly from any other treatment.

The PAI-1 antigen concentrations showed a median increase of 16.5% from baseline among women in the placebo group. Small declines were observed in all 3 active treatment arms, with the response in the 17-β estradiol–0.5-mg norethindrone acetate group (−7.0%) reaching statistical significance compared with placebo (P<.05).

Mean factor VII activity, expressed as a percentage of an age-adjusted normative value, did not change significantly from baseline in the placebo (−1.8%) or unopposed 17-β estradiol (1.4%) groups. The combined HRT groups showed changes from baseline in factor VII activity of −10.7% (17-β estradiol–0.25-mg norethindrone acetate) and −11.0% (17-β estradiol–0.5-mg norethindrone acetate), which were significantly different from the response in the unopposed 17-β estradiol group (P<.01 and P<.05, respectively), but not from that of the placebo arm. However, the comparison between 17-β estradiol–0.25-mg norethindrone acetate and placebo did approach significance (P<.06).

Antithrombin III activity, expressed as a percentage of an age-adjusted normative value, showed significant declines from baseline in all 4 treatment arms. The mean change among subjects in the placebo group was −9.7%. Each of the active treatment groups showed significantly larger declines vs placebo, but the responses did not differ between the unopposed 17-β estradiol (−17.0%; P<.01 vs placebo), 17-β estradiol–0.25-mg norethindrone acetate (−16.3%; P<.01 vs placebo), and 17-β estradiol–0.5-mg norethindrone acetate (−14.8%; P<.05 vs placebo) groups. Despite the marked reductions in antithrombin III activity in all active treatment groups, very few values dropped below the reference range during treatment. The number of subjects with values dropping below the reference range during treatment varied from 1 in the placebo and 17-β estradiol–0.5-mg norethindrone acetate arms to 4 in the 17-β estradiol and 17-β estradiol–0.25-mg norethindrone acetate groups.

Table 4 presents data on fasting glucose, HbA_1c, insulin, and C-peptide values at baseline, week 12, and end of treatment and changes from baseline to end of treatment according to group assignment. No significant differences between groups were present at baseline for any of these variables. Figure 4 displays the median percentage changes from baseline.

Fasting plasma glucose levels declined in all treatment groups, with mean changes ranging from −7.8% in the placebo arm to −10.4% in the unopposed 17-β estradiol arm. Responses were not significantly different between groups. Mean HbA_1c levels showed small changes from baseline ranging from 3.6% in the placebo group to −0.8% in the 17-β estradiol–0.5-mg norethindrone acetate group. The difference in HbA_1c response between placebo and 17-β estradiol–0.5-mg norethindrone acetate was the only comparison to reach statistical significance (P<.01).

Mean fasting insulin levels declined slightly in all treatment arms. The placebo group showed the smallest change from baseline (−2.8%), while the unopposed 17-β estradiol group had the largest response (−18.1%). Changes in the 17-β estradiol–0.25-mg and 0.5-mg norethindrone acetate groups were −11.1% and −6.2%, respectively. The response in the placebo group was significantly different from those of the unopposed 17-β estradiol (P<.01) and 17-β estradiol–0.25-mg norethindrone acetate (P<.05) groups. The responses in the 2 combined HRT arms did not differ from one another, although there was a significant difference between the responses of the unopposed 17-β estradiol and 17-β estradiol–0.5-mg norethindrone acetate groups (P<.05).

Mean fasting C-peptide concentrations showed a slight increase from baseline in the placebo group (2.0%). The unopposed 17-β estradiol (−12.6%%) and 17-β estradiol–0.25-mg norethindrone acetate (−7.8%%) groups had significant reductions in C-peptide concentrations compared with the placebo arm (P<.01 for both). Although the responses in the 2 combined HRT arms did not differ from one another, a significant difference existed between the responses of the unopposed 17-β estradiol and 17-β estradiol–0.5-mg norethindrone acetate groups (P<.01).

Baseline estradiol concentrations were similar in all treatment arms, with mean values ranging from 31 to 36 pmol/L (8.4 to 9.8 pg/mL) (P>.05 for all comparisons). At the end of the trial, all groups receiving hormones showed significant increases in estradiol concentration, with mean levels of 32, 291, 276, and 254 pmol/L (8.6, 79.3, 75.1, and 69.1 pg/mL) for the placebo, unopposed 17-β estradiol, 17-β estradiol–0.25-mg norethindrone acetate, and 17-β estradiol–0.5-mg norethindrone acetate groups, respectively (all P<.001 vs placebo).

Significant differences between treatments were noted for several adverse experiences, including breast pain, postmenopausal bleeding, leukorrhea, hot flushes, genital pruritus, and emotional lability. The numbers of women reporting breast pain were 5, 22, 26, and 34 in the placebo, unopposed 17-β estradiol, 17-β estradiol–0.25-mg norethindrone acetate, and 17-β estradiol–0.5-mg norethindrone acetate groups, respectively. The incidence of breast pain was increased vs placebo (P<.05) in each of the groups receiving active treatment, but did not differ between the active treatment arms. As expected, postmenopausal bleeding was more common in the active treatment arms than with placebo (P<.05), occurring in 5, 16, 23, and 23 of the women in the placebo, unopposed 17-β estradiol, 17-β estradiol–0.25-mg norethindrone acetate, and 17-β estradiol–0.5-mg norethindrone acetate groups, respectively. Similarly, leukorrhea was reported by 2, 16, 11, and 14 women, respectively (all P<.05 vs placebo). Genital pruritus was most commonly reported in the unopposed 17-β estradiol group (n = 8 vs 0-3 cases in the other groups; P<.05). Hot flushes (n = 8 in the placebo group vs 0-3 in the other groups; P<.05) and emotional lability (n = 4 in the placebo group vs 0 in the other treatment groups; P<.05) were most commonly reported in the placebo group.

Extensive data are available from clinical trials in which interventions used to lower LDL cholesterol and Apo B levels have reduced the frequency of cardiovascular events in women and men.^5,7,18- 20 In the present study, LDL cholesterol level was lowered by 10% to 14% from baseline in all of the groups receiving estrogen. Inclusion of norethindrone in the HRT regimen did not appear to interfere with the effects of 17-β estradiol on LDL cholesterol level. In fact, there was a nonsignificant trend toward enhanced LDL cholesterol reduction in the group receiving 17-β estradiol–0.5-mg norethindrone acetate. Unopposed 17-β estradiol, despite lowering LDL cholesterol level, had little effect on Apo B level and thus did not appear to reduce the number of circulating atherogenic lipoprotein particles. However, both groups receiving continuous combined HRT showed clinically important reductions (8%-9%) from baseline in Apo B concentration. Estrogens increase expression of hepatic LDL receptors, resulting in enhanced removal of LDL particles from the circulation.¹⁴ However, estrogens also increase the rate of very-low-density lipoprotein secretion.¹⁰ Therefore, the lack of net influence on Apo B by unopposed 17-β estradiol may reflect a reduction in LDL Apo B concentration that is offset by an increase in Apo B carried by very-low-density lipoprotein particles. Norethindrone may prevent the estrogen-induced increase in very-low-density lipoprotein secretion without altering estrogen's effect on LDL particle removal, resulting in a net decline in the circulating Apo B concentration. Other progestins, such as progesterone and 17-hydroxyprogesterone derivatives, have relatively little influence on very-low-density lipoprotein secretion and Apo B concentration.^5,11,21

Elevated levels of triglycerides and Lp(a), as well as depressed HDL cholesterol and Apo A-I levels, are associated with increased cardiovascular risk in both sexes.^22- 24 Moreover, the increase in relative risk of cardiovascular disease associated with hypertriglyceridemia is greater in women than men.^11,14 However, direct evidence from clinical trials to show that changes in these aspects of the lipid profile influence the cardiovascular event rate is limited.²⁵ In the present study, the increase in triglyceride levels induced by unopposed 17-β estradiol (9% increase from baseline%) was blunted by the inclusion of 0.25 mg of norethindrone acetate per day (5% increase from baseline) and prevented entirely by inclusion of 0.5 mg of norethindrone acetate (6% reduction from baseline) in the HRT regimen. Unopposed estrogen replacement raises triglyceride levels by increasing production of large, triglyceride-rich, very-low-density lipoprotein particles without producing a net increase in Apo B concentration.^5,11 Observational data suggest that triglyceride elevation in the absence of increased Apo B level may not increase cardiovascular risk.²³ Thus, the clinical importance of estrogen-induced hypertriglyceridemia and the ability of norethindrone to blunt or reverse this effect are of uncertain importance with regard to CHD risk.

Levels of HDL cholesterol showed an unexpected rise (9%) during the treatment period among women assigned to placebo, which could not be explained by changes in lifestyle habits, circulating estradiol concentrations, concomitant medication use, or nonnormality in the distribution of responses. Nevertheless, the mean concentration of Apo A-I, a major structural protein in HDL particles, showed a small decline from baseline (−1%) in the placebo group, suggesting that the elevation in HDL cholesterol concentration, if real, was not caused by an increase in the number of HDL particles. Increases in HDL cholesterol (14%) and Apo A-I (9%) levels were observed with unopposed 17-β estradiol. Norethindrone appears to blunt the 17-β estradiol–induced elevations in HDL cholesterol and Apo A-I levels in a dose-dependent manner.

Elevated levels of Lp(a) may increase CHD risk by enhancing both thrombogenesis and atherosclerosis.^26- 27 The Lp(a) concentration increased modestly (median change of 0.04 µmol/L [1.0 mg/dL] or 13%) from baseline among subjects in the placebo group but did not change in any of the active treatment arms. The finding that estrogen appeared to prevent the increase in Lp(a) observed among subjects taking placebo is consistent with results from previous studies in which reductions in Lp(a%) have been observed with ERT or HRT.¹² Norethindrone did not appear to modify this effect.

The degree to which cardiovascular disease risk may be influenced by the changes in the lipoprotein profile observed in this study is difficult to predict. Reductions in LDL cholesterol and Apo B levels would be expected to be protective, especially when accompanied by a reduction of triglyceride levels. The group receiving 17-β estradiol–0.5-mg norethindrone acetate showed the largest reductions in total cholesterol, LDL cholesterol, Apo B, and triglyceride levels, suggesting that norethindrone enhanced the favorable effects of 17-β estradiol on atherogenic lipoproteins. However, norethindrone also blunted or reversed the 17-β estradiol-induced increase in HDL cholesterol and Apo A-I levels, which might be expected to partially offset the beneficial changes observed in atherogenic lipoproteins.

The data from this study suggest that the net effect of 17-β estradiol plus norethindrone on circulating lipids and lipoproteins is favorable. Studies in cholesterol-fed rabbits that have undergone ovariectomy show that 17-β estradiol reduces the accumulation of aortic cholesterol and that norethindrone does not appear to counteract this benefit.³ Indeed, trends were observed toward enhanced protection in the animals receiving continuous 17-β estradiol plus norethindrone.¹³ However, firm conclusions will need to await data from clinical trials assessing the influence of these therapies on direct measures of atherosclerotic burden and/or cardiovascular events.

In the Heart and Estrogen-Progestin Replacement Study, thromboembolic events were 3 times more common in women receiving HRT than placebo, an excess of 4.1 per 1000 woman-years of observation.¹⁰ This finding is consistent with data from observational studies that have also reported increased risk of thrombosis among ERT or HRT users.^28- 29 The thrombogenic influence of HRT may also have played a role in the early increase in cardiovascular events observed among women assigned to the HRT arm in the Heart and Estrogen-Progestin Replacement Study.¹⁰ Therefore, comparing the effects of various HRT regimens on hemostatic variables may have important implications for understanding the influence of these therapies on ischemic and thrombotic events.

In the present study, subjects in the placebo group showed changes in hemostatic variables that would favor thrombogenesis, including increases in fibrinogen, PAI-1 antigen, and Lp(a) concentrations, as well as reduced antithrombin III activity. These may be attributable to seasonal variation, since most of the subjects were enrolled during the summer months (June to August) and completed the trial during the winter (December to March). Previous investigations have demonstrated that a relative thrombogenic state is present in the winter months among subjects living in climates similar to that in Chicago.^30- 31 Unopposed 17-β estradiol blunted or reversed the increases observed in fibrinogen and PAI-1 levels. These effects would be expected to reduce thrombogenicity. However, unopposed 17-β estradiol appeared to depress antithrombin III activity, which may enhance the risk of thrombus formation. However, the fact that antithrombin III activity in most of the women (>94% in all groups) remained within the normal range supports the prevailing view that reduced antithrombin III activity is not a major factor responsible for the increased risk of thromboembolic events associated with ERT or HRT use.^32- 33 Compared with unopposed 17-β estradiol, groups receiving 17-β estradiol plus norethindrone showed less blunting of the increase in fibrinogen concentration noted in the placebo group. Norethindrone had no apparent influence on PAI-1 concentration or antithrombin III activity, but it significantly depressed factor VII activity as compared with unopposed 17-β estradiol.

Data from the Heart and Estrogen-Progestin Replacement Study suggest that HRT can produce clinically important changes in hemostasis, resulting in excess thromboembolic events. Findings from the present trial highlight the fact that HRT regimens may vary with regard to thrombogenic potential, as indicated by differential effects on some markers of hemostatic function.

Some significant differences, although not large, between treatments were observed for changes from baseline in indicators of carbohydrate homeostasis. Unopposed 17-β estradiol produced significant reductions in the insulin and C-peptide levels, suggesting improved insulin sensitivity.³⁴ Insulin and C-peptide responses in the 17-β estradiol–0.25-mg norethindrone acetate group were intermediate compared with those in the unopposed 17-β estradiol and 17-β estradiol–0.5-mg norethindrone acetate groups. Norethindrone appeared to partially counteract the 17-β estradiol effects in a dose-dependent manner. A similar favorable effect of unopposed estrogen on fasting insulin and insulin resistance has been shown previously.^5,35 Progestins tend to blunt the estrogen-induced improvement in insulin resistance, as was observed in the present trial. Overall, 17-β estradiol plus norethindrone did not show any detrimental effect on carbohydrate metabolism compared with placebo; in fact, the HbA_1c level was reduced significantly more among subjects taking 17-β estradiol–0.5-mg norethindrone acetate than among those taking placebo. This is in accord with a previous report showing that 17-β estradiol–0.5-mg norethindrone acetate slightly reduced HbA_1c level and did not cause deterioration of insulin sensitivity, measured by the euglycemic, hyperinsulinemic clamp technique, in healthy postmenopausal women.³⁶

Postmenopausal hormone replacement affects many metabolic processes that have the potential to influence cardiovascular disease risk. Clinical trials provide evidence that HRT products may differ substantially with regard to their impact on the cardiovascular risk factor profile, depending on the specific estrogen and progestin used, as well as the dose, regimen, and route of administration. These differences may assist the medical practitioner in selecting an HRT product that is most compatible with the individual woman's cardiovascular disease risk profile. Results of the present investigation suggest that continuous combined HRT with 17-β estradiol, 1 mg, plus 0.25 to 0.50 mg of norethindrone acetate per day provides favorable changes in most of the biochemical markers of cardiovascular disease risk and has a profile distinctly different from that of unopposed 17-β estradiol, 1 mg. The overall impact on cardiovascular events of these different profiles needs to be investigated.

Accepted for publication April 28, 2000.

This research was funded by Novo Nordisk Pharmaceuticals, Inc, Princeton, NJ.

Reprints: Kevin C. Maki, PhD, Nutrition and Metabolism Research Unit, Chicago Center for Clinical Research, 515 N State St, 27th Floor, Chicago, IL 60610 (e-mail: kmaki@protocare.com).