Influence of Estrogen Therapy on Calcium, Phosphorus, and Other Regulatory Hormones in Postmenopausal Women: The MESA Study

Abstract

Background:

Estrogen therapy (ET) is associated with lower serum calcium and phosphorus concentrations and is known to increase bone mineral density (BMD). Other biomarkers of mineral metabolism may help understand the biological basis of these actions.

Methods:

We studied 2767 postmenopausal women in the Multi-Ethnic Study of Atherosclerosis, 862 (31%) of whom were using ET. We measured serum concentrations of calcium, phosphorus, 25-hydroxyvitamin D, 24,25-dihydoxyvitamin D, and fibroblast growth factor-23 and urinary fractional excretion of calcium (FEca) and phosphorus (FEphos). We examined the associations of ET with each biomarker. In addition, we tested whether the adjustment for biomarkers attenuated the association of ET with lumbar BMD measured by abdominal computed tomography in a subset of 810 women.

Results:

In adjusted models, women who used ET were younger in age [62 (SD 8) vs 66 (9) y, P < .001], had lower mean serum calcium [−13 mg/dL (95% confidence interval [CI] −0.17, −0.10), P < .001] and lower FEca [−0.15% (95% CI −0.21, −0.09), P < .001]. Mean serum phosphorus was lower [−0.19 mg/dL (95% CI −0.23, −0.15), P < .001] and FEphos [0.56% (95% CI 0.16, 0.96), P = .007] was higher in women on ET. Mean 25-hydroxyvitamin D and 24,25-dihydroxyvitamin D were higher [1.52 ng/dL (95% CI 0.57, 2.47), P = .002, and 0.26 ng/mL (95% CI 0.03, 0.48), P = .03, respectively] in women who used ET. Mean PTH and fibroblast growth factor-23 did not differ significantly by the use of ET. ET use was strongly associated with higher lumbar BMD [12.75 mg/cm³ (95% CI 7.77–17.73), P < .001]; however, mineral metabolism measures did not meaningfully alter this association.

Conclusions:

In a multiethnic cohort of postmenopausal women, ET use was associated with lower serum calcium, lower FEca, lower serum phosphorus, and higher FEphos, suggesting these associations are attributable to increased calcium intake into bone and increased urinary phosphorus excretion. ET use was also associated with greater concentrations of vitamin D metabolites. ET-associated differences in these mineral metabolism measures did not meaningfully attenuate the strong association between ET use and lumbar BMD.

Estrogen therapy (ET) is well known to induce osteoblast differentiation, inhibit osteoclasts, and promote bone deposition. Prior clinical studies demonstrate that ET prevents postmenopausal bone loss (1–4). ET also has important effects on calcium and phosphorus homeostasis, but the biological basis of these actions is poorly understood. Biomarkers of vitamin D and phosphorus metabolism have also been identified, including 24,25-dihdroxyvitamin D [24,25(OH)2D], a marker of vitamin D catabolism, and fibroblast growth factor-23 (FGF-23), a hormone that regulates urine phosphorous excretion. These and related biomarkers may help explain the effects of ET on mineral metabolism and bone deposition.

Prior studies evaluating ET effects on mineral metabolism have primarily focused on calcium, vitamin D, and PTH. Collectively these studies show that ET simultaneously lowers serum calcium and urine calcium excretion, resulting in a net influx of calcium into bone (5, 6). Natural menopause and associated declines in estrogen have the opposite effects. Menopause leads to a decline in 1,25 dihydroxyvitamin D [1,25(OH)2 vitamin D] and decreased calcium absorption from the intestinal tract (7–9), which may result in a modest increase in PTH (9). Additionally, prior work has predominantly evaluated Caucasian women, and recent evidence has suggested that the biological activity of mineral metabolism measures may vary by race (10).

Although phosphorus is a major component of bone mineral, much less is known about the effects of ET or menopause on phosphorus homeostasis and other mineral metabolites. We and others recently described that ET lowers serum phosphorus and simultaneously induces phosphaturia. Thus, the effects of ET on urine phosphorus handling may differ from calcium because ET lowers urine calcium excretion but appears to stimulate urine phosphorus excretion (11, 12). However, the relationship of ET on serum and urine phosphorus excretion remains unexplored in healthy, racially/ethnically diverse community-living women. Moreover, little is known about the relationship of ET with serum FGF-23 and 24,25(OH)2D.

24,25(OH)2D is an abundant metabolite that is formed when 25-hydroxyvitamin D [25(OH)D] is hydroxylated; however, the physiological effects of 24,25(OH)2D remain relatively unexplored. FGF-23 is a bone-derived hormone that induces urine phosphorus excretion, inhibits conversion of 25(OH)D to the active hormone 1,25(OH)2 vitamin D, and promotes degradation of vitamin D by stimulating the production of 24,25(OH)2D (13, 14). If ET induces decreases in FGF-23, it may provide a mechanism to explain how ET may induce greater 1,25(OH)2 vitamin D, which, in turn, may provide new insights to mechanisms through which ET promotes bone health.

In this study, we evaluated the association of ET with a panel of serum and urine markers of mineral metabolism in a large, well-characterized cohort of postmenopausal women in the Multi-Ethnic Study of Atherosclerosis (MESA). Based on pilot studies (11), we hypothesized that ET would be associated with lower serum phosphorus, greater urinary fractional excretion of phosphorus (FEphos), lower FGF-23, and greater 24,25(OH)2D. Lastly, we hypothesized that these changes may be part of the causal pathway linking ET with higher bone mineral density (BMD). Thus, we hypothesized that statistical adjustment for these measurements would at least partially attenuate the association of ET with BMD.

Materials and Methods

Study population

This was a cross-sectional study of the MESA study, a race/ethnically diverse cohort of middle-aged to older men and women from across the United States. Methods of the MESA study have been described in detail previously (15). Briefly, volunteers were recruited between July 2000 and August 2002 from six field centers around the United States (Baltimore, Maryland; St Paul, Minnesota; Chicago, Illinois; Forsyth County, North Carolina; New York, New York; and Los Angeles, California). The study population consists of 6814 men and women aged between 45 and 84 years of age, who are free of clinical cardiovascular disease (defined as myocardial infarction, angina, stroke, transient ischemic attack, heart failure, atrial fibrillation, use of nitroglycerin, prior angioplasty, coronary artery bypass grafting, valve replacement, pacemaker or defibrillator implantation, or any surgery on the heart or arteries) and identified themselves as non-Hispanic white, Chinese American, African American, or Hispanic. Institutional review boards at each site approved the study, and all participants granted written informed consent to participate. An ancillary study was funded to measure mineral metabolism markers using baseline samples on all study participants who had available stored serum (96% of the total MESA study population).

For this analysis, we limited the study population to women who had measures of mineral metabolism available at baseline (n = 3601). From these women, we excluded those who were premenopausal (n = 658) as determined from self-report in response to the questions, “Have you gone through menopause” or “Are you going through menopause?” We also excluded an additional 176 women who had missing covariate data, providing a final analytic sample of 2767 women. Of these, 810 women underwent measurement of lumbar BMD at either the second or third follow-up visit (∼18 and 36 mo after the baseline examination, respectively), which we evaluated as a subset for evaluating BMD.

Mineral metabolism measures

Markers of mineral metabolism measured at baseline included serum calcium, fractional excretion of calcium (FEca), serum phosphorus, FEphos, 25(OH)D, 24,25(OH)2D, PTH, and FGF-23. Blood and urine samples were obtained in the morning, after an overnight fast, and stored at −80°C. Serum calcium was measured by indirect ion selective electrode. Urine calcium and phosphorus levels were measured with a clinical chemistry analyzer (DXC600; Beckman Coulter). Urine calcium levels were determined by indirect potentiometry using a calcium ion-selective electrode and urine phosphorus concentrations by a timed end point method, (ie, inorganic phosphorus reacts with ammonium molybdate in an acidic environment to form a colored phosphomolybdate complex). For urine calcium, interassay coefficients of variation were 3.2% and 3.8% at 7.3 mg/dL and 10.2 mg/dL, respectively. For urine phosphate, interassay coefficients of variation were 3.4% and 2.8% at 25.6 mg/dL and 50.7 mg/dL, respectively. FEca was calculated from spot urine collections using the following formula: FEca = (urine calcium × serum creatinine × 100]/(serum calcium × urine creatinine]. Serum phosphorus was measured using a timed-rate colorimetry reaction. FEphos was calculated from spot urine collections using the following formula: FEphos = [urine phosphorus × serum creatinine × 100]/[serum phosphorus × urine creatinine]. Serum total 25(OH)D [sum of 25(OH)D2 and 25(OH)D3 and 24,25(OH)2D] was measured using HPLC-tandem mass spectrometry with internal standards at the University of Washington (16). Interassay coefficients of variation calculated using 81 repeat measurements of the quality control specimens placed in each plate of MESA samples were 8.5% at 24.8 ng/mL for 25(OH)D, 11.8% at 7.0 ng/mL for 25(OH)D, and 14.7% at 2.7 ng/mL for 24,25(OH)2D. Intact PTH concentration was measured in previously unthawed serum using the Beckman-Coulter DxI automated two-site immunoassay (Beckman-Coulter Inc). The interassay coefficient of variation was 6.1% at 30.1 pg/mL and 3.4% at 94.5 pg/mL. Serum intact FGF-23 was measured in 96-well plates using the Kainos sandwich immunoassay (Kainos), which measures the full-length (intact) FGF-23 molecule by recognizing both midmolecule and distal epitopes.

Lumbar BMD

An ancillary study was conducted to examine the calcification of the abdominal aorta. This ancillary study was not initiated until MESA examination 2 (∼18 mo after baseline), so abdominal computed tomography (CT) images were not available at the baseline examination concurrent with the mineral metabolism measures. Participants were randomly selected to undergo quantitative CT scanning at either the examination 2 or examination 3 follow-up visits. The CT scans were extended from the chest to the L2-L4 disc space, and the lumbar spine volumetric BMD was read (17). An electron-beam CT scanner (Chicago, New York City, and Los Angeles; Imatron C-150; General Electric Medical Systems) (18) or a multidetector CT system that used helical scanning with reconstruction in 5-mm-thick cuts and 350-mm field of view (New York, Forsyth County, and St Paul field centers; Siemens Inc, GE Medical Systems) was used. A previous study had demonstrated the comparability, accuracy, and reproducibility of these scanners (19). Participants were scanned along with phantoms of known physical calcium concentration to convert CT numbers directly to equivalent volumetric BMD in milligrams per cubic centimeter (20). CT data were collected using the Image Analysis QCT3DPLUS software program (Image Analysis) to determine BMD in a virtual 10-mm-thick slice of trabecular bone from the third lumbar vertebra. Scans were read centrally at the MESA Reading Center by a trained reader. In a random sample of 25, scans were reread on three occasions by the blinded scan reader; there was 100% agreement as to vertebrae data inclusion and no evidence of systematic differences between reads (21).

Covariates

Age, sex, race/ethnicity, and education level were self-reported. Smoking was categorized as never, former, or current use. History of cancer was determined by self-report. Season of the year (January-March, April-June, July-September, or October-December) corresponding to the collection of blood and urine samples was also recorded. Height and weight were measured with participants wearing light clothing (22). Body mass index (BMI) was calculated as weight/height² (kilograms per square meter). Study participants brought all study medications to the baseline examination. Medication use was recorded by trained study personnel, providing data on estrogen, thiazide, and loop diuretics. Only use of oral estrogens was considered ET.

The MESA Typical Week Physical Activity Survey was designed to identify the time spent in and frequency of various physical activities during a typical week in the past month (23). The MESA Typical Week Physical Activity Survey by design had the following summary measures: total hours per week and total metabolic equivalent (MET) minutes per week for the nine physical activity categories, three intensity levels, total physical activity, moderate and vigorous physical activity, and intentional exercise. We used the sum of moderate and vigorous physical activity in MET minutes per week.

Measurement of cystatin C was by a particle-enhanced immunonephelometric assay (N Latex Cystatin C; Siemens AG), and the intraassay coefficient of variation ranged from 2.0% to 2.8%. Estimated glomerular filtration rate (eGFR) was calculated using cystatin C in the formula: eGFR_tatin = 76.7 (cystatin C)^−1.19 (24). Urine albumin and creatinine were obtained from a single morning urine sample. Urine albumin was measured with nephelometry and urine creatinine was measured using the rate Jaffe method with a coefficient of variation of 2.5%–2.9%. Urine albumin to creatinine ratio (ACR) was expressed as milligrams per gram.

Statistical analysis

We compared baseline characteristics in ET- vs non-ET-using women using Student t tests, χ², or Mann-Whitney U tests as appropriate. Given right-skewed distributions, PTH and FGF-23 were log transformed for all analyses. We compared means in each mineral metabolism measure in women who did vs did not use ET using univariate and multivariate linear regression models, adjusting for age, race/ethnicity, education level (less than high school, high school graduate, greater than high school diploma), season of blood measurement, clinical site, current or former tobacco use, history of cancer, physical activity level (continuous METs per week), BMI, eGFR, urine ACR, and use of thiazide and loop diuretics. Similarly, we examined the adjusted relative percent difference in each mineral metabolism marker with estrogen use. We evaluated associations of mineral metabolism measures with BMD in univariate and multivariate linear regression, adjusting for the same covariates as above. Finally, we examined the association between ET and BMD alone, then by adding markers of mineral metabolism individually, and finally including all mineral metabolism measures concurrently to assess for attenuation. In all models, we tested for interactions by race/ethnicity and age.

Prior research has demonstrated that a major determinant of 24,25(OH)2D levels is 25(OH)D level (25–27). Thus, in further analyses, we investigated whether ET was associated with 24,25(OH)2D after additional adjustment for 25(OH)D.

Analyses were conducted using SPSS 16.0.2 and STATA 12.1. Values of P < .05 were considered statistically significant for all analyses including interaction terms.

Results

Among the 2767 women evaluated in this study, 862 (31%) were ET users. ET-using women were younger; were more likely to be white; achieved higher levels of education and income; were more likely to be former smokers; had lower BMI; were more physically active; and had higher eGFR, lower urine ACR, and higher lumbar BMD (Table 1).

Table 1.

Baseline Characteristics of Study Population (n = 2767)

	No Estrogen Use (n = 1905)	Estrogen Use (n = 862)	P Value
Mean (SD) age, y	66 (9)	62 (8)	.001
Race/ethnicity, %			<.001
White	31	56
Asian	14	7
Black	30	22
Hispanic	25	15
Educational achievement, %			<.001
Less than high school diploma	26	12
High school graduate	50	51
Greater than high school diploma	24	36
Income, %			<.001
<$20,000	37	19
$20,000–$39,999	30	27
$40,000–$74,999	21	28
≥$75,000	13	26
Mean (SD) BMI, kg/m2	28.7 (5.9)	27.0 (5.8)	<.001
Smoking status, %			<.001
Current	11	11
Former	26	39
History of cancer, %	9	8	.2
Moderate/vigorous physical activity, MET min/wk Monday–Sunday	3465 (1605, 6428)	3754 (1985, 6615)	.005
Mean (SD) eGFRtatin, mL/min per 1.73 m2	89 (19)	96 (18)	<.001
eGFR <60 mL/min per 1.73 m2, %	7	3	.001
HDL cholesterol	55 (14)	62 (17)	<.001
Median (IQR) urine ACR, mg/g	6.8 (4.3, 14.1)	5.0 (3.3, 8.7)	<.001
Use of thiazide diuretics, %	8	10	.04
Use of loop diuretics, %	2	2	.6
Mean (SD) lumbar BMD, mg/cm3	102 (37)	117 (38)	<.001
Season, %			.5
January–March	28	30
April–June	29	27
July–September	20	19
October–December	23	24
Site, %			<.001
Wake Forest University	13	21
Columbia University	21	8
Johns Hopkins University	14	18
University of Minnesota	17	16
Northwestern University	17	21
University of California, Los Angeles	21	18

Abbreviations: HDL, high-density lipoprotein; IQR, interquartile range.

In unadjusted analyses, mean calcium, FEca, serum phosphorus, PTH, and FGF-23 were lower among women who used ET, whereas 25(OH)D and 24,25(OH)2D were both higher (Table 2). FEphos levels did not differ between women who did vs did not use ET in unadjusted analyses. In multivariable adjusted models, women who used ET had lower mean serum calcium and lower FEca. Additionally, mean serum phosphorus was lower and FEphos was higher after multivariable adjustment in women on ET, indicating greater urinary phosphorus excretion (Table 2 and Figure 1). Mean 25(OH)D and 24,25(OH)2D were higher in women on ET in the multivariable models, albeit the associations were weaker compared with univariate models. Mean 24,25(OH)2D levels did not differ in women on ET after further adjustment for 25(OH)D. Mean PTH did not differ in women who did vs did not use ET after multivariable adjustment. Age, race, eGFR, and ACR were the covariates responsible for most of the attenuation of the association with PTH. Mean FGF-23 also did not differ in women who did vs did not use ET after multivariable adjustment (Table 2 and Figure 1); age, eGFR, and ACR were responsible for the greatest attenuation. In all models, results were similar irrespective of race/ethnicity or age (P interactions all >.05).

Table 2.

Means and Differences in Biomarkers of Mineral Metabolism Comparing Postmenopausal Women in MESA Who Were Using Estrogen With Women Who Were Not Using Estrogen Therapy (n = 2767)

	Unadjusted Means (95% CI)			Adjusted Means (95% CI)a
	No Estrogen	Estrogen	P Value for Difference	No Estrogen	Estrogen	P Value for Difference
Serum calcium, mg/dL	9.73 (9.71, 9.75)	9.60 (9.58, 9.63)	<.001	9.62 (9.37, 9.87)	9.48 (9.23, 9.74)	<.001
FEca, %	1.03 (1.00, 1.06)	0.90 (0.86, 0.94)	<.001	0.99 (0.54, 1.44)	0.85 (0.40, 1.30)	<.001
Serum phosphorus, mg/dL	3.91 (3.89, 3.93)	3.74 (3.71, 3.78)	<.001	4.35 (4.02, 4.67)	4.16 (3.84, 4.48)	<.001
FEphos, %	11.21 (10.98, 11.43)	11.31 (10.99, 11.63)	.6	9.64 (6.52, 12.75)	10.20 (7.09, 13.31)	.007
25(OH)D, ng/mL	21.85 (21.30, 22.40)	24.93 (24.14, 25.72)	<.001	28.42 (21.40, 35.44)	29.94 (22.92, 36.96)	.002
24,25(OH)2D, ng/mL	3.46 (3.33, 3.58)	4.17 (3.98, 4.35)	<.001	5.24 (3.59, 6.90)	5.50 (3.85, 7.15)	.03
PTH, pg/mL	47.01 (45.99, 48.04)	43.87 (42.52, 45.22)	<.001	42.44 (28.62, 56.26)	44.79 (29.98, 57.60)	.1
FGF-23, pg/mL	40.69 (39.93, 41.44)	38.73 (37.73, 39.72)	<.001	45.86 (35.66, 56.06)	45.01 (34.81, 55.20)	.2

Abbreviation: CI, confidence interval. Cell contents are mean values comparing women using estrogen therapy with those not using estrogen therapy, reported in units specific to each biomarker with 95% confidence interval.

^aAdjusted for age, race/ethnicity, education level, site, season, current or former tobacco use, history of cancer, physical activity level, BMI, eGFR, urine ACR, thiazide diuretics, and loop diuretics.

Figure 1.

Adjusted relative percentage difference in mineral metabolism markers with estrogen therapy use in postmenopausal women. Differences were adjusted for age, race/ethnicity, education level, site, season, current or former tobacco use, history of cancer, physical activity level, BMI, eGFR, urine albumin/creatinine, thiazide diuretics, loop diuretics. *, P < .05.

Next, we evaluated the association of each mineral metabolism measure with BMD in a subset of 810 women with available BMD measures (Table 3). Only lower FEca and higher 25(OH)D were associated with higher BMD in unadjusted models. After multivariable adjustment, these associations were attenuated. Race and age were responsible for most of the attenuation. In contrast, in the multivariable model, a significant direct association was revealed between FGF-23 and BMD.

Table 3.

Association Between Mineral Metabolism Markers and BMD in Women (n = 810)

	Univariate	Multivariatea
Serum calcium, per SD increase, mg/dL	0.75 (−2.03, 3.52)	1.63 (−0.74, 4.00)
FEca, per SD increase, %	−2.72 (−5.23, −0.22)b	−0.32 (−2.44, 1.81)
Serum phosphorus, per SD increase, mg/dL	−0.60 (−3.40, 2.20)	0.12 (−2.22, 2.48)
FEphos, per SD increase, %	−2.74 (−6.12, 0.63)	1.30 (−1.61, 4.22)
25(OH)D, per SD increase, ng/mL	3.22 (0.59, 5.85)b	−0.02 (−2.28, 2.24)
24,25(OH)2D, per SD increase, μg/mL	2.22 (−0.32, 4.77)	−0.34 (−2.49, 1.81)
Log PTH, per doubling, pg/mL	−0.53 (−5.08, 4.02)	−2.56 (−6.56, 1.45)
Log FGF-23, per doubling, pg/mL	−0.12 (−5.29, 5.06)	4.99 (0.51, 9.47)b

Cell contents are mean differences in BMD, reported in grams per cubic centimeter with 95% confidence interval.

^aAdjusted for age, race/ethnicity, education level, site, season, current or former tobacco use, history of cancer, physical activity level, BMI, eGFR, urine ACR, thiazide diuretics, loop diuretics, and estrogen therapy.

^bP < .05.

Of the 810 women with BMD measurements, 267 women (33%) were using ET. ET was strongly associated with higher BMD the multivariate model (Table 4). Addition of markers of mineral metabolism individually did not substantially attenuate this association. When all mineral metabolism measures were included in models simultaneously, ET remained strongly associated with higher BMD. These associations were similar, irrespective of race/ethnicity or age (P interactions all >.05).

Table 4.

Multivariate Association Between ET and BMD, With Adjustment for Mineral Metabolism Markers (n = 810)^a

	β-Coefficient	95% CI
ET	12.75	(7.77, 17.32)
ET + serum calcium	13.36	(8.32, 18.41)
ET + FEca	12.79	(7.78, 17.80)
ET + serum phosphorus	12.79	(7.74, 17.85)
ET + FEphos	12.70	(7.72, 17.69)
ET + 25(OH)D	12.53	(7.28, 17.79)
ET + 24,25(OH)2D	12.49	(7.24, 17.73)
ET + PTH	12.80	(7.82, 17.78)
ET + FGF-23	12.92	(7.95, 17.90)
ET + all mineral metabolism markers	13.17	(7.75, 18.59)

^aAdjusted for age, race, education level, site, season, current or former tobacco use, physical activity level, BMI, eGFR, urine ACR, thiazide diuretics, and loop diuretics.

The β coefficient represents the difference in BMD (in grams per cubic centimeter) comparing women using ET with those who are not using ET.

Discussion

In a large race/ethnically diverse community-living population of postmenopausal women, we found that use of ET was independently associated with lower serum calcium and FEca, suggesting greater deposition of calcium into bone, consistent with prior studies. We also observed that ET use was associated with simultaneously lower serum phosphorus and higher FEphos, suggesting that greater urinary phosphorus excretion lowers steady-state serum phosphate concentration with ET. ET was not associated with differences in PTH or FGF-23, which are known to regulate urine phosphorous excretion, supporting prior studies suggesting that estrogen may directly induce renal phosphorus excretion (12). Lastly, we observed that ET use was associated with higher 25(OH)D and 24,25(OH)2D. Among the multiple markers of mineral metabolism measured, only higher FGF-23 was associated with higher BMD in multivariable models. When examined individually and in composite, adjustment for the panel of mineral metabolism markers did not meaningfully attenuate the strong association between ET and BMD, suggesting that estrogen-associations in mineral metabolism markers are not likely to be on the causal pathway of ET and greater bone health.

Prior studies evaluating relationships of ET with mineral markers have evaluated mostly Caucasian populations and have primarily focused on serum and urine calcium, 25(OH)D, and PTH (5, 28–31). Our study confirms findings in these studies and extends them to other races/ethnicities. Prior studies have also noted decreased urinary calcium excretion with ET (30–32). In one study of 18 postmenopausal women, ET induced a 33% decrease in urinary calcium excretion (30). In our analysis, we found no interaction by race in the association between ET and calcium homeostasis. Our study also affirmed that among this diverse population of postmenopausal women, ET was simultaneously associated with lower serum phosphorus and higher FEphos, which may be due to estrogen's direct effect in down-regulating sodium phosphate cotransporters in the proximal tubule in the kidney, leading to renal phosphate wasting (6, 12, 33, 34). Prior studies evaluating this effect were conducted in laboratory animals or small clinical trials (32). Thus, our results extend the observations of ET on phosphorus homeostasis to a community-living, racially/ethnically diverse group of postmenopausal women.

We also confirmed that ET was associated with higher 25(OH) and 24,25(OH)2D. One previous study reported that a short course of ET led to higher 1,25(OH)2D levels in 10 postmenopausal women (5); however, the mechanism for the effect of ET on vitamin D metabolites remains unknown. A strength of our study was the ability to examine the association of ET and two vitamin D metabolites. In particular, the biological activity of 24,25(OH)2D has remained controversial. 24,25(OH)2D is formed with 24-hydroxylation of 25(OH)D and has traditionally been thought to be a degradation product. In vitro studies have reported that 24,25(OH)2D inhibits intestinal transport of calcium and phosphorus uptake in the intestine (25, 26) and phosphorus uptake in the kidney (35). Animal studies have also reported increases in bone mass when treated with 24,25(OH)2D (36, 37). Our findings extend the study of this less studied vitamin D metabolite to humans and show an association between ET and 24,25(OH)2D, which may be important in calcium and phosphorus regulation in humans. However, when we further adjusted for 25(OH)D, the association between ET use and 24,25(OH)2D was attenuated, which may suggest that most of the observed change in 24,25(OH)2D is related to its precursor, 25(OH)D, rather than direct effects on conversion of 24,25(OH)2D per se. Thus, our study suggests that ET may be associated with alterations in multiple biomarkers of mineral metabolism including not only phosphorus and calcium metabolism but also intermediate metabolites of vitamin D as well.

Few studies have examined the relationship between ET and FGF-23. We observed that ET was not associated with FGF-23 after adjustment for age, eGFR, and ACR. In contrast, in a prior study of older women with coronary heart disease, we reported that ET was independently associated with lower FGF-23 levels (11). The results from the current analysis were based on a large population of middle-aged to older women from a diverse range of racial/ethnic backgrounds and excluded individuals with cardiovascular disease. Differences in study designs may explain the differences in results between these two studies, particularly because we found that age and level of kidney function were important confounders in the association between ET and FGF-23 in the present study.

In regard to the relationships of mineral metabolism markers with BMD, we observed that only higher FGF-23 was associated with higher BMD. It is well established that excess FGF-23 due to hereditary disorders such as hypophosphatemic rickets leads to phosphate wasting and impaired bone deposition in children and adults. Although some animal studies have suggested that elevated FGF-23 may directly contribute to defects in bone deposition (38, 39), this has not been consistently observed in humans. For example, a study among 2800 Swedish men found no association of FGF-23 with BMD (40). A recent analysis of 77 obese perimenopausal women reported an inverse association of FGF-23 with BMD (41). Two other studies, one conducted among 50 obese postmenopausal women and the other among elderly men, reported no association between FGF-23 and BMD (40, 41). Although the exact mechanism to explain the direct association between FGF-23 and BMD in our study is unclear, our results build on prior clinical investigations by studying this mineral marker in a relatively healthy and diverse cohort of postmenopausal women.

Lastly, we found that statistical adjustment for the available markers of mineral metabolism did not statistically attenuate the association between ET and BMD. Thus, our results suggest that the direct effects of ET on osteoclasts and osteoblasts may be the predominant mechanisms linking ET to greater BMD (42). Alternatively, indirect mechanisms other than mineral metabolism may play an important role. It has been suggested that estrogen is intimately involved in bone health by inhibiting osteoclasts and inhibiting bone resorption.

Strengths of our study include evaluation of a large, well-characterized cohort of postmenopausal women, its multiethnic nature, and evaluation of women from six centers across the United States. The numerous available markers of mineral metabolism is another key strength. Measures of kidney function and many other potential confounding variables were available concurrently. The study also has important limitations. We did not have information about the indication, type, duration, age at start, or dose of ET. BMD was measured by quantitative CT and we did not have other measures of BMD such as dual-energy x-ray absorptiometry scans. We did not have measures of BMD at anatomic sites other than the lumbar spine; BMD measures were available only in a subgroup and then 18–36 months after the mineral metabolism markers. Unfortunately, 1,25(OH)2D and markers of bone resorption were not measured in this cohort and may have provided additional insights. We did not have detailed information on calcium and vitamin D supplementation among study participants. This was a cross-sectional observational study; thus, causality of these associations could not be determined. It remains unknown whether ET affects longitudinal trajectories in mineral metabolism markers and whether trajectories of change in mineral markers may influence BMD.

In conclusion, in a large, multiethnic community-living population of postmenopausal women, we demonstrate that ET is associated with lower calcium and lower FEca; lower serum phosphorus and higher FEphos; and higher 25(OH)D and higher 24,25(OH)2D. Among the mineral metabolism measures, only FGF-23 is associated with higher BMD. Adjustment for mineral metabolism markers in isolation or collectively did not statistically influence the association of ET with lumbar spine BMD, suggesting that ET-induced alterations in mineral metabolism is not on the causal pathway linking ET use and higher BMD. Thus, further studies are needed to elucidate the complex regulation and physiological effects of ET on mineral metabolism and bone health and to determine whether ET-associated changes in mineral metabolism are related to other health-related outcomes in postmenopausal women.