Effects of Estrogen with Micronized Progesterone on Cortical and Trabecular Bone Mass and Microstructure in Recently Postmenopausal Women

Abstract

Context:

In women, cortical bone mass decreases significantly at menopause. By contrast, loss of trabecular bone begins in the third decade and accelerates after menopause.

Objective:

The aim of the study was to investigate the effects of estrogen on cortical and trabecular bone.

Design:

The Kronos Early Estrogen Prevention Study is a double-blind, randomized, placebo-controlled trial of menopausal hormone treatment (MHT) in women, enrolled within 6–36 months of their final menstrual period.

Setting:

The study was conducted at the Mayo Clinic, Rochester, Minnesota.

Intervention:

Subjects were treated with placebo (n = 31), or .45 mg/d conjugated equine estrogens (n = 20), or transdermal 50 μg/d 17β-estradiol (n = 25) with pulsed micronized progesterone.

Main Outcome Measures:

Cortical and trabecular microarchitecture at the distal radius was assessed by high-resolution peripheral quantitative computed tomography.

Results:

At the distal radius, cortical volumetric bone mineral density (vBMD) decreased, and cortical porosity increased in the placebo group; MHT prevented these changes. By contrast, MHT did not prevent decreases in trabecular microarchitecture at the radius. However, MHT prevented decreases in trabecular vBMD at the thoracic spine (assessed in a subset of subjects; n = 51). These results indicate that MHT prevents deterioration in radial cortical vBMD and porosity in recently menopausal women.

Conclusion:

The maintenance of cortical bone in response to estrogen likely has important clinical implications because cortical bone morphology plays an important role in bone strength. However, effects of MHT on trabecular bone at the radius differ from those at the thoracic spine. Underlying mechanisms for these site-specific effects of MHT on cortical vs trabecular bone require further investigation.

Since Fuller Albright (1) established the relationship between estrogen (E) deficiency and postmenopausal osteoporosis, many studies have demonstrated that E treatment initiated after menopause prevents bone loss and fractures (2–5). Based on these early studies, E deficiency was identified as the major cause of the early, accelerated, and slow phases of bone loss in postmenopausal women (6). This paradigm has been modified (7) based on significant advances in imaging techniques that have allowed insights into the patterns of changes in specific skeletal compartments throughout the lifetime. In a longitudinal population-based study (8), the rate of bone loss from the axial skeleton (predominantly comprised of trabecular bone) in women was similar before and after 50 years of age, whereas bone loss at appendicular sites (predominantly comprised of cortical bone) did not begin until midlife. Additional cross-sectional (9) and longitudinal (10) population-based studies using quantitative computed tomography (QCT) to assess bone geometry and volumetric bone mineral density (vBMD) at the lumbar spine, femoral neck, distal radius, and distal tibia confirmed that cortical vBMD remains relatively stable until menopause and then decreases significantly. By contrast, trabecular bone loss begins early, in the third decade, and continues at an accelerated rate after menopause (9, 10). These observational data suggest that loss of cortical bone is closely tied to E deficiency, whereas trabecular bone loss may be due to intrinsic age-related processes that are modulated by E deficiency. However, this hypothesis has not been directly tested because virtually all clinical trials of E treatment have used dual-energy x-ray absorptiometry (DXA), which cannot separate trabecular from cortical bone or assess bone microarchitecture.

Therefore, this study evaluated the effects of menopausal hormone treatment (MHT) on bone microarchitecture of trabecular and cortical bone at the distal radius using high-resolution peripheral QCT (HRpQCT) in recently menopausal women enrolled in the Kronos Early Estrogen Prevention Study (KEEPS) at Mayo Clinic (Rochester, Minnesota). HRpQCT is a 3-dimensional imaging technique that provides measurements of bone macrostructure (eg, cortical thickness and vBMD) of cortical and trabecular bone separately, as well as bone microarchitecture (eg, trabecular number/connectivity and cortical porosity). A better understanding of hormonal regulation of cortical vs trabecular bone compartments will provide new insights into the underlying mechanisms responsible for bone fragility in women after menopause.

Subjects and Methods

Study subjects

Women enrolled in KEEPS (NCT00154180) at Mayo Clinic participated in this substudy. KEEPS was a multicenter, double-blinded, 4-year prospective randomized controlled trial to test the hypothesis that MHT initiated early in menopause reduced progression of atherosclerosis as defined by increases in carotid intimal medial thickening and coronary arterial calcification (11, 12). Women were enrolled within 6–36 months of their final menstrual period and randomized to 0.45 mg/d of oral conjugated equine estrogens (CEE), 50 μg/d of transdermal 17β-estradiol (both with oral, micronized progesterone, 200 mg for 12 d each month), or placebo. Complete rationale for the study design and for the selection of drugs and dosages has been described in detail previously (11). Menopause was defined as a serum FSH level ≥ 35 ng/mL and/or estradiol levels < 40 pg/mL. Exclusions included a history of malignancies, myocardial infarction, angina, congestive heart failure, thromboembolic disease, heavy smoking (>10 cigarettes/day), morbid obesity (body mass index > 35 kg/m²), dyslipidemia (low-density lipoprotein cholesterol > 190 mg/dL), hypertriglyceridemia (triglycerides > 400 mg/dL), uncontrolled hypertension (systolic blood pressure > 150 mm Hg and/or diastolic blood pressure > 95 mm Hg), and fasting glucose > 126 mg/dL. No participant was using specific bone-active medications.

A CONSORT flowchart describes the progress of the subset of KEEPS participants at Mayo Clinic throughout the 4-year trial (Figure 1). The present analysis was performed on the subset of KEEPS participants who completed HRpQCT measurements and all 4 years of treatment (adherence to treatment sample; n = 76). All personnel were blinded to study treatment. The study was approved by the Mayo Clinic Institutional Review Board. Participants provided informed written consent.

Figure 1.

CONSORT flow diagram describing the enrollment, randomization, and progress of the KEEPS participants through the trial by randomization group at the Mayo Clinic site. Complete follow-up and inclusion in the ITT (intention to treat) analysis is defined as completion of the primary endpoint (HRpQCT) and acceptable scan quality. Inclusion in the secondary adherence to treatment analysis is further defined as completion of 4 years of treatment.

HRpQCT measurements

Detailed descriptions of the HRpQCT device and in vivo imaging processing and analysis protocol used at Mayo Clinic have been reported (13–17) and are summarized here briefly. The Xtreme CT (Scanco Medical AG, Brüttisellen, Switzerland) was used to assess cortical and trabecular bone microstructure and vBMD at distal sites of the nondominant radius. A scout scan of the forearm was performed to view the intersection of the joint space with midjoint lines of radius where the reference line was appropriately placed. The scanner was programmed to subsequently acquire a fixed 3-dimensional stack of 110 high-resolution computed tomography (CT) slices starting at 9.5 mm proximal from the reference line and extending 9.02 mm proximally, with an isotropic voxel size and slice thickness of 82 μm. Total scan time was 2.8 min, with an effective energy of 40 keV (radiation dose of 65 mrad), a field of view of 125.9 mm, and an image matrix of 1536 × 1536 pixels. Trabecular bone volume fraction (bone volume/tissue volume [BV/TV], %) was derived from the average trabecular vBMD (mg/cm³) within the trabecular region, assuming that fully mineralized bone has a matrix mineral density of 1200 mg HA/cm³, whereas the marrow background is equivalent to 0 mg HA/cm³. A thickness-independent structure extraction was used to identify 3-dimensional ridges (center points of the trabeculae) because individual trabeculae will not be resolved at their correct thickness due to partial volume effects (14). Consequently, trabecular number (TbN, 1/mm) was taken as the inverse of the mean spacing of the ridges (15). Analogous with histomorphometry (18), trabecular thickness (1/mm) was derived as (BV/TV)/TbN, and trabecular separation (mm) was derived as (1 − BV/TV)/TbN. The validity of this approach has been rigorously tested, and excellent correlations (r ≥ .96) have been shown between HRpQCT and the “gold standard” ex vivo micro-CT technique (16).

For the cortical parameters, the cortex was segmented from the grayscale image with a Gaussian filter and threshold (15). Recognizing that the default cortical bone analysis provided with this HRpQCT device performs poorly for subjects with thin or porous cortices (19, 20), we used the extended cortical analysis available from the manufacturer to obtain cortical vBMD (mg/cm³), cortical area (mm²), cortical thickness (1/mm), endocortical circumference (mm), and periosteal circumference (mm). Furthermore, we derived cortical pore volume (CtPoV; mm³) and cortical porosity (CtPo; %) using a validated approach (21) that has been used by several groups (22, 23), including our own (17). CtPoV is a direct voxel-based measure of the volume of the intracortical pore space, whereas CtPo is a relative voxel-based measure of the volume of the intracortical pore space normalized by the sum of the pore and cortical bone volume. Short-term precision (coefficients of variation) of the HRpQCT device in our laboratory has been reported previously (13), based on repeat measures on 20 volunteers on the same day after repositioning.

Thoracic spine trabecular vBMD measurements

As previously described (24, 25), a 64-slice detector CT scanner (Siemens Sensation 64; Siemens Medical Solutions, Forchheim, Germany) was used to assess coronary artery calcification at baseline and year 4. The thoracic spine was also captured within the CT scan region of interest. Trabecular vBMD of the T8 and T9 vertebrae was determined from these images using the Spine Cancer Assessment program (Biomedical Imaging Resources, Mayo Clinic) (26). Strong associations have been shown between vBMD measurements of the thoracic and lumbar spine (25, 27). Each vertebra was manually identified, extracted, and reoriented so that the spinal processes were horizontal in the transverse image and the body was vertical in the coronal and sagittal orientations (25). Then, in a standard coronal image, a 3-point cursor was used to select the appropriate measurement plane by placing 2 points horizontally on the superior endplate and a third point on the inferior endplate. The spinal processes were then removed from the analysis using an elliptical mask that was manually positioned on the central transverse slice and applied to all sections. The oblique section was computed from the original image volume using 3-dimensional cubic interpolation (25). Voxel intensities were linearly scaled to a calibration phantom (Mindways Software, Austin, Texas) placed in the region of interest, such that the resulting image represented the calcium content in milligrams per cubic centimeter at each voxel.

Thoracic trabecular vBMD was computed from the center slice of the T8 and T9 vertebrae. However, due to partial volume effects and the thin vertebral cortex (28), this segmentation technique tends to overestimate the vBMD of the resulting trabecular zone (25). Recognizing this limitation, thoracic trabecular vBMD was also assessed from a central zone (25), which excluded any trabecular bone near the cortex. Thus, a 4-year percentage change in thoracic spine vBMD was assessed at four different vertebra sites (ie, the trabecular and central zones of T8 and T9). Baseline and 4-year thoracic spine vBMD measurements were only available on 51 (placebo = 22, oral E = 16, transdermal E = 13) of the 76 participants who had acceptable baseline and 4-year HRpQCT scans and had completed all 4 years of treatment.

DXA measurements

Areal bone mineral density (aBMD; g/cm²) was assessed from DXA scans performed on the nondominant radius (ultradistal and total), nondominant hip (femoral neck trochanter, diaphysis, and total), and lumbar spine (L1–L4) (Lunar Prodigy System; GE Healthcare, Madison, Wisconsin).

Statistical analyses

All anthropometric and bone imaging data are presented as mean ± SE. Differences among the 3 groups in baseline descriptive and HRpQCT characteristics of the radius were tested using ANOVA followed by least significant difference post hoc test. The primary analysis of percentage change in HRpQCT-derived trabecular and cortical bone parameters from baseline to 4-year follow-up was performed on the adherence to treatment sample (n = 76) described earlier. Intent-to-treat (ITT) analyses (n = 86) are reported in Supplemental Tables 1–3 (published on The Endocrine Society's Journals Online web site at http://jcem.endojournals.org). Changes in cortical and trabecular bone parameters assessed by HRpQCT were similar in the oral and transdermal E groups (see Table 2), and thus were combined for all subsequent analyses. The paired samples t test was used to compare percentage change (95% confidence interval) in bone parameters from baseline to 4 years in the placebo- and hormone-treated groups. Furthermore, comparisons are reported in percentage change over the 4 years of the study in bone parameters derived from DXA, HRpQCT, and CT between the placebo- and hormone-treated groups using the unpaired (independent sample) t test. Analyses were performed using The Statistical Package for the Social Sciences for Windows, version 19.0 (SPSS, Chicago, Illinois). All testing was performed at a significance level of P < .05 (2-tailed).

Table 2.

Comparisons of Percentage Change in HRpQCT Parameters From Baseline to Year 4 Between Participants on Oral vs Transdermal E Treatment

Radius HRpQCT	Oral E (n = 20)	Transdermal E (n = 25)	P
Cortical vBMD, mg/cm3	−0.8 ± 1.0	−1.0 ± 0.7	.903
Cortical porosity, %	2.5 ± 8.9	14.0 ± 7.4	.322
Cortical pore volume, mm3	1.6 ± 8.2	16.4 ± 7.1	.180
Cortical area, mm2	−0.9 ± 4.1	1.8 ± 1.9	.527
Cortical thickness, mm	−1.5 ± 4.1	2.1 ± 2.4	.431
Endocortical circumference, mm	0.4 ± 2.1	−0.9 ± 1.5	.606
Periosteal circumference, mm	0.1 ± 1.6	−0.4 ± 1.1	.813
Trabecular BV/TV	−7.9 ± 2.2	−5.7 ± 1.7	.431
Trabecular area, mm2	0.7 ± 4.0	−1.9 ± 3.0	.598
Trabecular number, 1/mm	−3.5 ± 2.5	0.0 ± 2.0	.277
Trabecular thickness, mm	−4.6 ± 1.9	−5.6 ± 1.4	.676
Trabecular separation, mm	4.4 ± 2.7	0.7 ± 2.1	.279

Values are presented as mean ± SE.

Results

At baseline, age, anthropometric characteristics, and cortical and trabecular bone parameters at the distal radius were not significantly different among the groups (Table 1). The mean percentage change from baseline to 4 years in cortical and trabecular bone parameters at the distal radius was not different between the 2 treatment groups (Table 2).

Table 1.

Baseline Descriptive and Radius HRpQCT Characteristics of the KEEPS Participants

	Placebo (n = 31)	Oral E (n = 20)	Transdermal E (n = 25)	P
Descriptives
Age, y	52.6 ± .4	53.6 ± .3	53.1 ± .5	.274
Total menopausal age, mo	17.3 ± 1.8	17.3 ± 2.0	20.1 ± 2.0	.541
Height, cm	167 ± .9	164 ± 1.2	166 ± 1.1	.198
Weight, kg	75.1 ± 2.1	75.3 ± 2.5	70.8 ± 2.3	.292
Body mass index, kg/m2	27.1 ± .7	28.1 ± 1.0	25.5 ± .8	.106
Radius HRpQCT
Cortical vBMD, mg/cm3	1013 ± 6.8	1031 ± 6.9	1038 ± 10.0	.175
Cortical porosity, %	0.84 ± .07	0.98 ± .08	0.88 ± .12	.543
Cortical pore volume, mm3	4.4 ± .38	5.4 ± .47	4.9 ± .63	.374
Cortical area, mm2	57.2 ± 1.6	58.8 ± 1.3	61.7 ± 1.9	.159
Cortical thickness, mm	1.03 ± .04	1.11 ± .03	1.13 ± .04	.178
Endocortical circumference, mm	47.4 ± 1.0	44.2 ± 1.3	45.0 ± .9	.139
Periosteal circumference, mm	65.4 ± 1.0	62.3 ± 1.4	63.6 ± .9	.083
Trabecular BV/TV	0.115 ± .005	0.121 ± .005	0.122 ± .005	.504
Trabecular area, mm2	181 ± 7.0	158 ± 9.9	163 ± 6.6	.079
Trabecular number, 1/mm	1.69 ± .05	1.70 ± .05	1.63 ± .06	.599
Trabecular thickness, mm	0.068 ± .002	0.072 ± .002	0.076 ± .002	.118
Trabecular separation, mm	0.54 ± .02	0.53 ± .02	0.56 ± .03	.515

Values are presented as mean ± SE.

The 4-year percentage changes in cortical and trabecular bone parameters at the distal radius are shown in Figure 2. Cortical vBMD at the distal radius decreased significantly in the placebo group (−2.9% [−3.9 to −2.0%]; P < .001), whereas MHT prevented the loss of cortical vBMD over 4 years (−0.9% [−2.1 to +.3%]; P > .05). Furthermore, the change in cortical vBMD at the distal radius between placebo- and hormone-treated groups was statistically significant (difference = 2.0%; P = .015) (Table 3). CtPo at the distal radius increased significantly in the placebo group (+29.5% [+17.1 to +41.9%]; P < .001), whereas MHT prevented the increase in CtPo (+8.9% [−2.7 to +20.4%]; P > .05). The difference in change of CtPo at the distal radius between placebo- and hormone-treated groups was significant (difference = 20.6%; P = .018) (Table 3). Similarly, the difference in change of CtPoV at the distal radius was also significant (Table 3). There were no significant differences in changes in cortical area, cortical thickness, endocortical circumference, and periosteal circumference of the distal radius between the placebo- and hormone-treated groups (Table 3).

Figure 2.

Percentage change in cortical (A) and trabecular (B) bone parameters at the distal radius in placebo compared with E-treated subjects. Data are shown as mean (±SE) of the percentage change after 4 years of treatment. *P < .05 for change from baseline to 4 years.

Table 3.

Comparisons of Percentage Change in Radius HRpQCT and DXA Parameters From Baseline to Year 4 Between Participants on Placebo vs E Treatment

	Placebo (n = 31)	Estrogen (n = 45)	P
Radius HRpQCT
Cortical vBMD, mg/cm3	−2.9 ± 0.5	−0.9 ± .6	.015
Cortical porosity, %	29.5 ± 6.1	8.9 ± 5.7	.018
Cortical pore volume, mm3	28.8 ± 6.1	9.8 ± 5.4	.025
Cortical area, mm2	−1.3 ± 1.3	0.6 ± 2.1	.494
Cortical thickness, mm	−0.6 ± 1.9	0.5 ± 2.2	.719
Endocortical circumference, mm	0.2 ± 1.3	−0.3 ± 1.2	.786
Periosteal circumference, mm	2.1 ± 2.4	−0.2 ± 0.9	.313
Trabecular BV/TV	−10.7 ± 1.6	−6.7 ± 1.4	.066
Trabecular area, mm2	0.3 ± 2.7	−0.7 ± 2.4	.783
Trabecular number, 1/mm	−4.1 ± 1.4	−1.6 ± 1.6	.248
Trabecular thickness, mm	−6.6 ± 1.5	−5.2 ± 1.1	.432
Trabecular separation, mm	5.4 ± 1.5	2.4 ± 1.7	.207
DXA regional aBMD
Femoral neck, g/cm2	−4.5 ± 0.7	−0.9 ± .8	.003
Trochantar, g/cm2	−2.4 ± 1.3	−0.7 ± 1.1	.318
Diaphysis, g/cm2	−3.8 ± 0.7	−1.0 ± 0.8	.015
Total hip, g/cm2	−3.4 ± 0.8	−0.5 ± 0.8	.020
Lumbar spine L1–L4, g/cm2	−3.7 ± 0.9	3.6 ± 0.7	<.001
Ultradistal radius, g/cm2	−5.8 ± 1.2	−0.7 ± 0.9	<.001
Total radius, g/cm2	−4.3 ± 0.8	−0.5 ± 0.5	<.001

Values are presented as mean ± SE.

In contrast, MHT did not prevent deterioration of the trabecular bone parameters at the distal radius (all P values < .001), including trabecular BV/TV (E = −10.7% [−14.0 to −7.4%]; placebo = −6.7% [−9.5 to 3.9%]) and trabecular thickness (E = −5.2% [−7.4 to −2.9%]; placebo = −6.6% [−9.6 to −3.6%]) (Figure 2). There were no significant differences in any of the 4-year percentage changes in any of the other trabecular bone parameters at the distal radius between the placebo- and hormone-treated groups (Table 3).

Despite the differential responses of cortical vs trabecular bone to MHT observed at the distal radius using HRpQCT, MHT consistently prevented loss of aBMD assessed by DXA at the hip, lumbar spine, and radius as compared with placebo treatment (Table 3). The mean percentage change in aBMD ranged from −7.1 to −2.4% across skeletal regions for the placebo group as compared with −2.0 to +3.5% for the hormone-treated group.

The changes in CT measures of trabecular vBMD at the four thoracic spine sites (trabecular and central zones of the T8 and T9 vertebrae) are shown in Figure 3. The average of the percentage changes (95% confidence interval) in trabecular vBMD across all thoracic spine sites increased slightly in the hormone-treated group (+1.7% [−1.4 to +4.9%]; P > .05), whereas it decreased significantly in the placebo group (−9.0% [−12.2 to −5.2%]; P < .001). The changes in trabecular vBMD between the placebo- and hormone-treated groups were significant (all P values < .001) at all thoracic spine sites.

Figure 3.

Percentage change in thoracic spine vBMD (mg/cm³) at the trabecular (A) and central (B) zones of the T8 and T9 vertebrae in placebo compared with E-treated subjects. Data are shown as mean (±SE) of the percentage change after 4 years of treatment. **P < .01; ***P < .001 for change from baseline to 4 years.

Discussion

In this randomized placebo-controlled trial of 4 years of continuous treatment with either CEE or transdermal 17-β estradiol plus pulsed progesterone in recently postmenopausal women, MHT prevented cortical, but not trabecular, bone loss at the distal radius. In contrast, MHT prevented the loss of trabecular vBMD at the thoracic spine and aBMD at the hip, lumbar spine, and radius. The findings at the distal radius are consistent with cross-sectional (9) and longitudinal (10) observational findings and support current paradigms, suggesting that loss of cortical bone is closely tied to E deficiency, whereas trabecular bone loss is largely due to intrinsic age-related processes independent of E (7).

DXA is a valuable clinical tool to assess overall bone health, but it does not allow separate analysis of trabecular from cortical bone and cannot detect independent changes in these underlying skeletal compartments. The advantage of HRpQCT is that such separate analysis can be performed, and as indicated by the present study, both oral and transdermal E treatments similarly reduced cortical bone loss at the distal radius. These findings extend those of previous studies, demonstrating that administration of transdermal E can be as effective as oral therapy in the prevention of postmenopausal bone loss (29, 30). Transdermal delivery of E is less likely to cause adverse procoagulant effects associated with oral E (31), which may lower the risk for deep venous thrombosis. Thus, given the equal efficacy of the 2 modes of E delivery in the preservation of radial cortical bone, transdermal E may be a better option for concomitant prevention of bone loss and fractures with reduced risk for adverse thrombotic events.

The maintenance of cortical bone at the distal radius in response to MHT during the early postmenopausal years likely has important implications for fracture prevention later in life. Indeed, the mechanical competence of a bone (its ability to sustain a certain load without undergoing fracture) is dependent not only on the mass and intrinsic properties of bone tissue, but also on the bone's morphology (geometric and microarchitectural arrangement of trabecular and cortical bone) (32). Moreover, cortical bone morphology likely plays a particularly important role in fracture prevention for a number of reasons. First, cortical bone comprises over 80% of the adult skeleton (33) and is increasingly recognized for its critical structural role in bone strength and fracture resistance (34, 35). Second, 80% of all fractures that occur after the age of 65 occur at nonvertebral sites that are predominantly comprised of cortical bone (36). Third, most of bone loss that occurs at appendicular sites after the age of 65 is cortical (37). Lastly, haversian canals transverse the cortex, creating an intracortical surface area that exposes the cortex and leaves it vulnerable to the higher bone remodeling rates associated with E deficiency after menopause (38). The resulting erosion of cortical bone results in increased cortical porosity, which has a destabilizing effect on the structural integrity of the bone, comprising its strength and resistance to fracture. This phenomenon was observed in the present study because recently postmenopausal women taking placebo treatment experienced a 30% increase in cortical porosity at the distal radius over 4 years. This increase may be clinically significant because the peak stress at yield of cortical bone is greatly reduced when cortical porosity increases by only a few percent (39, 40).

An unexpected finding of the present study was a differential response of MHT on trabecular vBMD at the radius as compared with the spine. One explanation is that single-energy CT measures of vertebral vBMD are affected by changes in the ratio of red to yellow marrow (41). This is relevant because E treatment has been shown to prevent increases in bone marrow adipocytes in postmenopausal osteoporotic women (42). Thus, the effect of MHT on changes in vertebral trabecular vBMD assessed by single-energy CT may be confounded because E treatment decreases the ratio of yellow to red marrow at this site, which in turn, falsely elevates the measurement of trabecular vBMD. This may explain why there was a slight increase (+1.7%) in thoracic spine trabecular vBMD in the MHT groups over 4 years. In contrast to the spine, this is not an issue at the distal radius because bone marrow is completely yellow at this site from young adulthood onward (43). For these reasons, measures of changes in cortical and trabecular bone at the distal radius may be more reliable than those at the spine.

In addition to altering measurement of vBMD, the red marrow of the spine is rich in immune cells that produce proinflammatory cytokines (eg, IL-1β, IL-6, TNFα, granulocyte-monocyte-colony stimulating factor, macrophage colony-stimulating factor, and prostaglandin E₂), which support osteoclast differentiation, activity, and survival (44, 45). With E deficiency, there is an increase in proinflammatory cytokines (45), whereas this effect is reversed by E treatment (46). To the extent to which these act locally vs systemically, E treatment may prevent bone resorption at the spine, but not at the distal radius.

Yet another potential explanation for the differences in trabecular bone response to MHT at the radius and spine may be a result of dose-response and/or site-specific effects. In a 2-year longitudinal study, 0.6 mg/d of CEE prevented vertebral trabecular vBMD loss in postmenopausal women, whereas lower doses (0.15 or 0.3 mg/d of CEE) were ineffective, although those doses fully protected against cortical bone loss at the forearm (4). Relatively higher doses of E may be necessary to prevent trabecular bone loss at the spine but are ineffective at preventing trabecular bone loss at the radius. Future dose-response trials using advanced imaging techniques are warranted to determine the minimal effective dose of E necessary to prevent trabecular bone loss at various skeletal sites.

In contrast to the lack of E effect on trabecular bone at the distal radius, our study demonstrated that 0.45 mg/d of oral CEE and 50 μg/d of transdermal 17β-estradiol prevented the decrease in cortical vBMD (3%) and the increase in cortical porosity (30%) observed in the placebo group. These findings suggest that at the distal radius, cortical bone may be more responsive than trabecular bone to lower doses of E. Unfortunately, due to the limited spatial resolution of single-energy CT and the thin vertebral cortex (28), an accurate determination of the cortical bone at the spine is not possible. Thus, it was not possible to test whether E treatment had a differential effect on cortical vs trabecular bone at the thoracic spine.

The vast majority of E intervention studies in postmenopausal women have been performed using a dose of E (ie, 0.625 mg/d of CEE, equivalent to 1.0 mg/d of 17β-estradiol), thought for many years to be the “lowest effective dose” (2–5). These studies provide concrete evidence that this dose and formulation of E treatment prevents bone loss and reduces the risk of fracture among postmenopausal women (2–5, 47). However, given the results of the Women's Health Initiative (48), current clinical recommendations are for lower doses of E. Consequently, there has been considerable interest in lower doses of E therapy for the prevention of postmenopausal bone loss and fractures (49), and thus far, a number of studies have demonstrated that doses that keep serum estradiol levels well below the premenopausal range have beneficial effects on bone. For example, in a 2-year longitudinal study comparing 0.3, 0.45, and 0.625 mg/d of CEE vs placebo in recently (mean age, 52 y) postmenopausal women, all 3 doses were effective in preventing bone loss assessed by DXA and reducing bone turnover (50). Furthermore, 3 years of low-dose E treatment with 17β-estradiol (0.25 mg/d) increased bone mineral density of the hip, lumbar spine, and total body and reduced bone turnover in healthy older (>65 y of age at enrollment) postmenopausal women (51). Importantly, this study found that this lower dose of oral E was not associated with adverse thrombotic events. More recently, an ultra-low dose of E (ie, 0.0125 mg/d of 17β-estradiol) had similar effects on bone turnover and the preservation of bone (52). Notwithstanding these promising results, the optimal dose of E for the maintenance of bone and prevention of fractures and other safety endpoints (eg, breast cancer, stroke, cardiovascular disease, deep venous thrombosis, and dementia) remains to be defined. Additional analysis of KEEPS participants may address some of these additional questions. The present study showed that a lower dose of E affected cortical but not trabecular bone compartments in the wrist and may suggest that the optimal dose of E may someday be individualized based on a composite risk assessment that includes more detailed bone structural analysis.

The strengths of the KEEPS trial include its randomized 4-year prospective design, comparison of 2 different E delivery routes (oral and transdermal), enrollment of women soon after the onset of menopause (within 6–36 mo), and use of state-of-the-art bone imaging (HRpQCT), which allows for separation of trabecular and cortical bone compartments and accurate assessments of bone microarchitecture. There are, however, some limitations. First, cortical and trabecular bone compartments were only assessed at the distal radius; due to the limited spatial resolution of single-energy CT and the thin vertebral cortex (28), an accurate determination of the cortical bone at the spine is not possible. Furthermore, bone parameters at the hip were measured by DXA. Thus, we were not able to test whether the results we observed at the radius for cortical bone were consistent at the spine and hip. We acknowledge that the changes in cortical bone we observed at the distal radius may not necessarily reflect the changes in cortical bone at other skeletal sites. Second, baseline and 4-year thoracic spine vBMD measurements were not available on 25 of the women included in HRpQCT analysis. Thus, we acknowledge that this may have influenced the thoracic spine results. Lastly, because the HRpQCT scans were only performed at the Mayo Clinic site, this was a fairly small sample.

In conclusion, this randomized controlled trial demonstrates that 4 years of either transdermal or oral E plus pulsed micronized progesterone treatment leads to the preservation of cortical bone at the distal radius in recently postmenopausal women. In contrast, these treatments did not prevent trabecular bone loss at the distal radius after the onset of menopause. MHT was able to prevent loss of trabecular bone at the thoracic spine. Further animal and human studies are needed to define the underlying mechanisms for the differential effects of MHT on cortical vs trabecular bone as well as for the site-specific differences in the responses of trabecular bone.