|Home | About | Journals | Submit | Contact Us | Français|
To investigate the effect of aromatase inhibitors (AI) on intestinal calcium absorption, measured using the gold-standard dual stable calcium isotope method. In this pilot study, we recruited 10 postmenopausal women with hormone receptor-positive breast cancer who planned to initiate AI therapy; women receiving chemotherapy were excluded. Women completed two 24 h inpatient calcium absorption study visits, the first prior to AI therapy and the second at least 6 weeks following onset of AI therapy. We calculated total fractional calcium absorption (TFCA) using the dose-corrected fractional recovery of two stable isotopes from 24 h urine collections. Ten postmenopausal women (mean ± SD age, 66 ± 7 years; 25(OH)D 40 ± 7 ng/mL, and total calcium intake of 1,714 ± 640 mg/day) exhibited no change in TFCA related to AI therapy (0.155 ± 0.042 prior to and 0.160 ± 0.064 following AI therapy, p = 1.0). Subjects exhibited a surprisingly small decline in serum estradiol levels with AI therapy that was not statistically significant. However, there was a significant correlation between duration of AI therapy and the decline in serum estradiol levels (r = −0.65, p = 0.040). In this pilot study, AI therapy did not decrease TFCA. Women with early stage breast cancer exhibited an unexpectedly low TFCA, most likely due to their high calcium intake. The null effect of AI therapy on TFCA might relate to the brief duration of AI therapy, the minimal effect of AI therapy on estradiol levels, subjects’ high calcium intake or excellent vitamin D status.
In postmenopausal women with hormone receptor-positive breast cancer, physicians initiate aromatase inhibitors (AI) either as initial adjuvant therapy or following a course of tamoxifen. Compared with tamoxifen, AIs improve disease- free and overall survival in women with breast cancer [1, 2]. Based on these results, the American Society of Clinical Oncology clinical practice guidelines recommend that clinicians consider AIs for all postmenopausal women with hormone receptor-positive breast cancer . However, unlike tamoxifen which has estrogen-like effects on the skeleton, AI therapy causes bone loss . Postmenopausal women with breast cancer on AI therapy have an increased fracture risk compared to their peers [5–7].
AIs inhibit conversion of androgens to estrogens, reportedly reducing serum estrogen levels by 81–96% within 2–6 weeks of starting therapy [8, 9]. Presumably through low estrogen levels, AI therapy associates with increased bone turnover, a decline in bone mineral density, and an increased risk of fracture [4, 7, 10]. To our knowledge, no studies have investigated whether AI-induced declines in tissue estrogen levels lead to a decline in intestinal calcium absorption.
Menopause and subsequent reduced estrogen levels result in decreased intestinal calcium absorption. Researchers recruited 189 women ages 35–45 years old and studied longitudinal changes in calcium absorption over two decades, reporting a decline in calcium absorption with age (~ 0.2% per year), and an additional one time decrease in calcium absorption (~ 2.2%) at the time of menopause . In another study , women entering menopause experienced a sudden decline in calcium absorption that corresponded to a decline in serum estradiol levels, whereas women with ongoing menses experienced stable calcium absorption. In a third study , women experienced a significant decline in calcium absorption following oophorectomy, which was reversed with estrogen therapy. In summary, several studies report that menopause results in a decrease in calcium absorption efficiency.
Since AIs decrease serum estrogen levels, and menopause with associated estrogen deficiency results in decreased calcium absorption efficiency, we hypothesized that AIs might also decrease intestinal calcium absorption. If true, decreased calcium absorption might contribute to AI-associated declines in bone mineral density and a commensurate increase in risk of fracture. We conducted a pilot study to determine the feasibility of testing this hypothesis in breast cancer patients starting AI therapy.
We recruited postmenopausal women with recently diagnosed hormone receptor-positive breast cancer who planned to initiate AI therapy following definitive local therapy (lumpectomy or mastectomy, and/or local radiation therapy). The primary study outcome was the change in subjects’ total fractional calcium absorption (TFCA)  following at least 6 weeks of AI therapy. Secondary study outcomes included changes in bone turnover markers, Short-Form 12 survey scores, and the incidence of musculoskeletal signs and symptoms related to AI therapy. The University of Wisconsin Human Subjects Committee approved the study protocol and all participants provided written informed consent prior to study procedures. The study was registered as a clinical trial (ClinicalTrials.gov identifier: NCT00766532).
TFCA is a continuous variable with a between-subject standard deviation of 1% and a within-subject precision error of TFCA of <1%, among postmenopausal women studied in our center . Although researchers have previously reported a ~ 2% decline in TFCA related to menopause , such small changes in TFCA have not been linked to relevant health outcomes such as a higher fracture risk. By contrast, in the Study of Osteoporotic Fractures , 5,452 women underwent a single measurement of calcium absorption and subsequently, researchers noted that an 8% decrease in fractional calcium absorption corresponded to an increased risk of hip fracture among these women. We therefore chose to calculate sample size for this pilot study based on findings of the Study of Osteoporotic Fractures. Recruitment of 10 women and use of the paired t test allowed us to detect an 8% or greater difference in TFCA within subjects after AI therapy with ~ 99% power and a two-sided type 1 error of 0.05. Our ability to detect an 8% or greater difference in TFCA with AI therapy was deemed clinically relevant to the risk of fracture in women initiating aromatase inhibitor therapy.
All data was graphed and analyzed using the Shapiro–Wilk test for normality. Subjects’ demographic, laboratory and dietary characteristics were summarized using the mean and standard deviation, with the addition of the median and interquartile range for the few variables without a normal distribution. Within-subject changes in TFCA, laboratory and dietary variables were analyzed using the paired t test for normally distributed data, or the Wilcoxon test for data without a normal distribution, with its use noted next to the relevant p value. Analyse-It software (Leeds, UK) was used for all analyses.
We recruited women with recently diagnosed hormone receptor-positive breast cancer who planned to start AI therapy following definitive local therapy (lumpectomy, mastectomy, and/or local radiation therapy). Eligible women were also required to be at least 5 years past the onset of menopause, defined as the date of last menses or bilateral oophorectomy. We excluded women undergoing chemotherapy for breast cancer, due to its potential independent effects on intestinal health, calcium homeostasis, and bone turnover. We excluded women with intestinal conditions associated with malabsorption, including Crohn disease, ulcerative colitis, pernicious anemia, bacterial overgrowth, celiac sprue, chronic diarrhea, or use of antibiotics within the month preceding the screening visit. We also excluded women with known Stage 4 or 5 chronic kidney disease, defined as an estimated GFR <30 mL/min, as our calcium absorption method is not validated in this population. We excluded women taking medications known to interfere with calcium or vitamin D metabolism, including oral glucocorticoids or anticonvulsants.
Oncologists (AT, MEB, and KW) referred potential participants to our research center shortly after their diagnosis of breast cancer, and prior to onset of AI therapy. Recruitment began in January 2009 and ended in June 2010. Twenty-eight women with breast cancer were interviewed by phone (Fig. 1). We met women who appeared eligible based on questions administered by phone, and obtained written informed consent. Eligible subjects underwent two 24 h inpatient TFCA study visits, the first study serving as a baseline measurement prior to onset of AI therapy. Baseline TFCA measurements permitted us to control for multiple patient-specific variables that influence TFCA [17, 18] including age, race, calcium intake, and serum vitamin D levels. Based on studies showing that estrogen levels decrease by 81–96% from baseline levels approximately 2–6 weeks after onset of AI therapy [8, 9], we required at least 6 weeks of AI therapy before the second TFCA study. We considered the 6-week interval to be sufficient to detect the full effect of estrogen suppression on calcium homeostasis, based on guidance from the Institute of Medicine .
We measured TFCA using two stable calcium isotopes, which are calcium compounds of different atomic weights with no known toxicity . The isotopic distribution of calcium in the environment is 96.9% 40calcium, 2.1% 44calcium, 0.6% 42calcium, with minor amounts of calcium with atomic weights of 43, 46, and 48. We purchased purified 44calcium and 42calcium from Trace Sciences International (Wilmington, DE) as calcium carbonate powder. The University of Wisconsin Waisman Clinical Biomanufacturing Facility prepared the calcium isotope solutions for this study. Personnel dissolved the calcium powders in HCl, mixed the resulting solution with NaCl and adjusted the solution pH 6 for intravenous injection or pH 4 for oral administration. Solutions were sterile filtered using a Sterivex GV 0.22 micron sterile filter, aseptically dispensed into sterile vials and tested for sterility and pyrogenicity. We administered one isotope (44calcium) orally to trace intestinal TFCA and administered the other isotope (42calcium) intravenously to trace renal recycling and intestinal calcium secretion. The ratio of the two isotopes in a 24 h urine collection provides an accurate estimate of TFCA and is considered the current gold-standard method for measuring calcium absorption . The Wisconsin State Lab of Hygiene-Trace Element Research Group, under the direction of Dr. Martin Shafer, measured the calcium isotope content of urine samples using highresolution mass spectrometry as described elsewhere [15, 18], with a ratio standard deviation of 0.2–0.5% for each sample.
Subjects fasted from midnight until 0700 when they arrived on the research unit to undergo measurement of calcium absorption. Subjects provided a urine sample upon arrival on the research unit, representing the second morning void; this sample was used to measure urine N-telopeptide levels. A morning blood sample was collected for measurement of calcium, creatinine, 25(OH)D, 1,25(OH)2D, intact parathyroid hormone, estradiol, bonespecific alkaline phosphatase, and C-reactive protein. Serum estradiol levels were measured by chemiluminescent microparticle immunoassay (Architect assay, Abbott Laboratories) with a lower limit of detection of 7 pg/mL and a between-run coefficient of variation of 4.9% for measurement of levels in the low range, defined as a mean value of 45 pg/mL.
With breakfast, women consumed a glass of calciumfortified orange juice (≤50 mL) containing one stable calcium isotope tracer (8 mg of 44Ca). At the same time, the nurse injected another stable calcium isotope tracer (3 mg of 42Ca) intravenously, flushing the line with saline and weighing the syringe before and after use to record the dose of 42Ca. Over the next 24-h period, we collected all urine to calculate TFCA based on the dose-corrected fractional recovery of the two stable calcium isotopes established by Eastell et al.  using the formula shown below:
During each TFCA study, the study dietitian (LAD) adjusted the breakfast meal to ensure that subjects consumed 300 mg of dietary calcium from breakfast. Breakfast was consumed while subjects received oral and intravenous stable calcium isotope tracers. Including breakfast, the mean ± SD dietary calcium intake during the first and second TFCA studies was 1,090 ± 290 mg, and 1,070 ± 260 mg, respectively. During each TFCA study, women also continued to take their outpatient supplements, including calcium. Subjects consumed three meals and one or two snacks with approximately 17% of caloric intake from protein, 38% from fat, and 45% from carbohydrates. Food provided during the first and second TFCA study was identical.
Subjects completed two 4 day diet diaries, one immediately after the first TFCA study and one just before or after the second TFCA study. The research dietitian taught subjects to complete accurate diet diaries, and each subject received a scale to weigh food portions. Subjects recorded what they ate every other day for 4 days including 1 weekend day, to capture daily variability in nutrient intake. Diet diaries were analyzed using Food Processor® Nutrition Analysis Software, ESHA Research (Salem, OR). We used data collected from diet diaries as covariates when determining factors influencing TFCA.
At each TFCA visit, subjects rated their degree of morning stiffness, pain (10-point visual analog scale) and underwent a joint examination by a rheumatologist (KEH) to determine the number of tender and swollen joints among 28 joints. In addition, subjects completed the Short-Form 12 survey version 1 as a means of assessing mental and physical health.
Twelve postmenopausal women with early stage breast cancer provided written informed consent to take part in the study (Fig. 1). Two women subsequently withdrew from the study, one due to inability to establish intravenous access, and the other due to a personal decision to defer AI therapy.
Ten women (nine Caucasian and one African American) with a mean ± SD age of 66 ± 7 years old completed all study visits. All 10 women underwent lumpectomy and 8 women were also treated with local radiation therapy. Women began AI therapy (anastrozole 1 mg by mouth daily) immediately after completing their first TFCA study and took AI therapy an average of 53 days (range 42–90 days) between their first and second TFCA study visits. Self-reported adherence to anastrozole was excellent; no subject reported interruption of therapy. Unlike two other studies [8, 9], subjects in our study exhibited a surprisingly small decline in serum estradiol levels with AI therapy that was not statistically significant (Table 1). However, there was a significant correlation between duration of AI therapy and the decline in serum estradiol levels (r = −0.65, p = 0.040).
TFCA did not change as a result of AI therapy. TFCA was 0.155 ± 0.042 at baseline, and 0.160 ± 0.064 following the onset of AI therapy (p = 1.000). There was no significant relationship between the absolute change in fractional calcium absorption and duration of AI therapy (r = −0.04, p = 0.907). We found no significant relationship between serum estradiol levels and TFCA at baseline (r = −0.04, p = 0.905) or following AI therapy (r = −0.49, p = 0.146). Likewise, there was no significant relationship between AI-induced declines in estradiol levels and change in fractional calcium absorption (r = 0.06, p = 0.873). Finally, we analyzed changes in TFCA as a function of detectable (n = 5) or undetectable (n = 5) estradiol levels following AI therapy. Among subjects with undetectable estradiol levels, TFCA increased by 2.2%. By contrast, TFCA decreased by 1% in subjects with measurable estradiol levels, a between-group change that was not statistically significant (p = 0.55, Wilcoxon test).
We next analyzed changes in milligrams of calcium absorbed per day (TFCA times total calcium intake from diet and supplements). Follow-up measurements for milligrams of calcium absorbed were non-parametric, and therefore all data are presented as the median (interquartile range). Subjects absorbed 259 mg (190–308 mg) of calcium at baseline and 255 mg (118–328 mg) of calcium following AI therapy (p = 1.000, Wilcoxon test). There was no significant relationship between the absolute change in milligrams of calcium absorbed and duration of AI therapy (r = −0.07, p = 0.852). We also found no significant relationship between serum estradiol levels and milligrams of calcium absorbed at baseline (r = −0.43, p = 0.218) or following AI therapy (r = −0.15, p = 0.679). Finally, there was no significant relationship between AI-induced declines in estradiol levels and the change in milligrams of calcium absorbed per day (r = 0.11, p = 0.761).
Table 1 summarizes patients’ laboratory results and dietary habits at each inpatient study visit. Six variables demonstrated a non-normal distribution and are also summarized using the median (interquartile range). Of note, baseline serum 25(OH)D levels were >30 ng/mL in 9 of 10 subjects and the remaining subject had a serum 25(OH)D level of 29.7 ng/mL, indicating vitamin D repletion in all subjects . During the study, subjects exhibited no significant changes in parameters of calcium homeostasis including bone turnover markers. Subjects’ 4 day food records detected no significant changes in dietary habits between study visits. Most importantly, subjects did not report significant changes in their use of supplemental calcium or their consumption of dietary calcium.
Participants reported mild pain that did not change during the study (pain level 1.2 ± 1.5 before, and 1.4 ± 1.4 following onset of AI therapy, p = 0.910). One woman had rheumatoid arthritis; her joint pain, tender joint count, swollen joint count and duration of morning stiffness did not change with AI therapy. Aside from this individual, no other subject reported morning stiffness. No subject experienced new joint pain or joint swelling during AI therapy.
Subjects completed the SF12 survey during each calcium absorption study visit. The physical component summary scale remained unchanged during the study (48.5 ± 10.3 prior to and 48.4 ± 10.0 following AI therapy, p = 1.000). Similarly, the mental component summary scale did not change during the study (54.2 ± 5.2 prior to and 53.5 ± 6.7 following AI therapy, p = 0.508).
Postmenopausal women who receive AI therapy to treat breast cancer can experience a decline in bone mineral density and a commensurate increased risk of fracture [5–7]. We hypothesized that because calcium absorption decreases in the setting of estrogen deficiency, that AI therapy might decrease calcium absorption, thereby contributing to a greater risk of osteoporosis and fracture. In this pilot study, we found no significant decline in calcium absorption resulting from 6 weeks of AI therapy.
We found an unexpectedly low TFCA among subjects in this study, compared to postmenopausal women participating in other studies in our center [15, 18]. We believe the observed low TFCA among subjects in this study relates to subjects’ high calcium intake, as several other researchers [11, 14, 17, 22] report an inverse relationship between calcium intake and TFCA.
Possible explanations for the null effect of AI therapy on calcium absorption in this study include the short duration of AI therapy, the null effect of AI therapy on estradiol levels, the high calcium intake among participants, and participants’ excellent vitamin D status. Excellent vitamin D status might reduce the effects of AI therapy on calcium absorption and/or bone mineral density. In early stage breast cancer patients randomized to exemestane, researchers  found a trend toward greater rates of spine and hip bone loss among subjects with baseline serum 25(OH)D levels <30 ng/mL, compared to subjects with baseline levels >30 ng/mL. In a subsequent review article, Geisler and Lonning  emphasized calcium and vitamin D supplements as one strategy to manage bone loss related to AI therapy. It is possible that the excellent vitamin D status of subjects in the current study helped to minimize any negative effect of AI therapy on TFCA or bone turnover.
As joint pain is a common side effect of AI therapy, we systematically monitored joint pain and swelling during the conduct of this pilot study. None of our subjects reported joint pain or were noted to have new synovitis when examined by a rheumatologist. The lack of new synovitis is not surprising, as the median time to onset of initial joint pain in the “Arimidex, Tamoxifen, Alone or In Combination” (ATAC) Trial was 14 months . In addition, our subjects did not have prior chemotherapy, which might be a risk factor for AI-induced joint pain .
Our study had several key strengths, including the use of a gold-standard method to measure calcium absorption, and performance of calcium absorption studies on the inpatient unit, permitting complete 24 h urine collections to measure calcium absorption. We used a paired study design in which each subject served as her own control, thereby controlling for patient-specific factors that influence calcium absorption including age, race, habitual calcium intake, and vitamin D status. In terms of potential weaknesses, our sample size was small and therefore we had limited power to detect small changes in fractional calcium absorption. We measured changes in calcium absorption after 6 weeks of AI therapy; it is possible that the true effect of AI therapy on calcium absorption might take longer than 6 weeks, especially since subjects did not demonstrate a significant decline in estradiol levels despite excellent self-reported adherence. The precision of the clinical estradiol assay used in this study might not have captured the true effect of AI therapy on estradiol levels in our subjects. In addition, it is possible that the excellent vitamin D status of our subjects might have prevented AI-induced declines in TFCA. We did not restrict or standardize calcium intake, and therefore cannot determine from this study whether breast cancer itself might cause, or be associated with, low TFCA.
In this pilot study, subjects with early stage breast cancer did not experience a decrease in TFCA following 6 weeks of AI therapy. Women with early stage breast cancer exhibited an unexpectedly low TFCA, most likely due to their high calcium intake. The null effect of AI therapy on TFCA might relate to the brief duration of AI therapy, the null effect of AI therapy on estradiol levels, subjects’ high calcium intake or excellent vitamin D status. We suggest that future studies assessing the influence of AI therapy on calcium absorption recruit a larger sample of subjects, standardize subjects’ calcium intake and vitamin D status, and prescribe AI therapy for a longer duration in order to better evaluate whether AI therapy truly decreases TFCA.
We thank the participants, who made this research possible. We thank the University of Wisconsin Clinical Research Unit nursing staff for outstanding research assistance and patient care. KEH received salary support from the National Institute of Health (K23 AR050995 and R01 AG028739) during the conduct of this study. The project was supported by grants from the American College of Rheumatology/Research Education Foundation with the American Society for Specialty Physicians through the Hartford Foundation and Atlantic Philanthropies (Junior Career Development Award in Geriatric Medicine) and by grant UL1RR025011 from the Clinical and Translational Science Award (CTSA) Program of the National Center for Research Resources, National Institutes of Health.
Conflict of interest The authors declare that they have no conflicts of interest.
Amye Tevaarwerk, University of Wisconsin School of Medicine and Public Health, Room 4124, 1685 Highland Avenue, Madison, WI 53705-2281, USA. University of Wisconsin Carbone Cancer Center, WIMR 6037, 1111 Highland Ave., Madison, WI 53705, USA.
Mark E. Burkard, University of Wisconsin School of Medicine and Public Health, Room 4124, 1685 Highland Avenue, Madison, WI 53705-2281, USA. University of Wisconsin Carbone Cancer Center, WIMR 6059, 1111 Highland Ave., Madison, WI 53705, USA.
Kari B. Wisinski, University of Wisconsin School of Medicine and Public Health, Room 4124, 1685 Highland Avenue, Madison, WI 53705-2281, USA. University of Wisconsin Carbone Cancer Center, WIMR 6033, 1111 Highland Ave., Madison, WI 53705, USA.
Martin M. Shafer, Wisconsin State Lab of Hygiene, Madison, WI, USA.
Lisa A. Davis, University of Wisconsin Institute for Clinical and Translational Clinical Research, Madison, WI, USA.
Jyothi Gogineni, Centegra Health System, Crystal Lake, IL, USA.
Elizabeth Crone, University of Wisconsin School of Medicine and Public Health, Room 4124, 1685 Highland Avenue, Madison, WI 53705-2281, USA.
Karen E. Hansen, University of Wisconsin School of Medicine and Public Health, Room 4124, 1685 Highland Avenue, Madison, WI 53705-2281, USA.