Calcified Tissue International

, Volume 88, Issue 3, pp 223–230

Does Anastrozole Affect Bone Resorption Similarly in Early and Late Postmenopausal Women?

Authors

    • Charles Salt Centre for Human MetabolismRobert Jones & Agnes Hunt Orthopaedic & District Hospital NHS Trust
  • Richard A. Cochrane
    • Breast Unit, Wrexham Maelor Hospital
  • Michael W. J. Davie
    • Charles Salt Centre for Human MetabolismRobert Jones & Agnes Hunt Orthopaedic & District Hospital NHS Trust
Original Research

DOI: 10.1007/s00223-010-9452-2

Cite this article as:
Powell, D.E., Cochrane, R.A. & Davie, M.W.J. Calcif Tissue Int (2011) 88: 223. doi:10.1007/s00223-010-9452-2

Abstract

The aim of this study was to determine whether the bone-resorption response to anastrozole differed according to initial patient age in postmenopausal women with breast cancer in a cross-sectional study. Second-morning void urines were collected for measurement of urinary cross-linked N-telopeptide of type I collagen (uNTx, corrected for creatinine and log-transformed) from postmenopausal women, 99 with breast cancer on anastrozole (ABC), 88 with newly diagnosed breast cancer (NDBC), and 137 community-dwelling healthy control (HC) women. Bone mineral density (BMD) was also measured at the lumbar spine (LS, L2–L4) and the femoral neck (FN) in the ABC group. uNTx (nanomole bone collagen equivalents/millimole creatinine) levels increased with age in HC subjects. In patients <70 years, anastrozole treatment led to a significant increase in uNTx compared with age-related HC subjects (1.74 vs. 1.55, P < 0.005). Patients >70 years showed no such increase compared to HC (1.72 vs. 1.69, nonsignificant); however, NDBC women >70 years had uNTx levels significantly lower than HC women (1.59 vs. 1.69, P < 0.05). There was no difference in uNTx levels above and below the age of 70 years in NDBC women (1.56 vs. 1.59, nonsignificant). ABC women were more likely to have a positive LS BMD z score than age-matched controls. Anastrozole treatment increases bone turnover more in younger postmenopausal women with breast cancer than in older women compared to healthy controls. Higher LS BMD in ABC patients may help protect against fracture.

Keywords

Aromatase inhibitorElderlyBone resorptionBreast cancerPostmenopausal

Adjuvant treatment of breast cancer using third-generation aromatase inhibitors has yielded significant improvements in increasing disease-free survival and reducing the reoccurrence of cancer [1, 2]. The success of this therapeutic approach has been tempered by drawbacks including a greater incidence of clinical fracture [1, 2]. By inhibiting the aromatase enzyme, circulating levels of estradiol can be lowered by as much as 10-fold by these agents [35]. The reduction of estradiol levels following menopause is held to be responsible for the decline in bone mineral density (BMD) and for an increase in bone-resorption markers and fracture risk [69]. The further reduction of postmenopausal estradiol levels with aromatase inhibitors is thought to lead to higher levels of bone-resorption markers [10, 11].

Although values of bone-resorption markers, such as urinary cross-linked N-telopeptide of type I collagen (uNTx), tend to increase with age [12, 13], the effect of age on the most estrogen-sensitive bone, which is largely trabecular and found in the distal radius and vertebrae [14], is controversial. Using QCT, bone loss at the lumbar spine appears to be constant postmenopausally [15], whereas using dual-energy X-ray absorptiometry (DXA, which measures the whole vertebra) bone loss at the spine is confined to a period between the middle of the fifth decade and the end of the seventh decade [16].

In breast cancer studies the reported increase in bone-resorption marker response to anastrozole is variable. Only two previous studies have looked at uNTx in postmenopausal women with breast cancer treated with anastrozole. In the Anastrozole, Tamoxifen, Alone or in Combination (ATAC) trial uNTx levels increased by an average of 15% from baseline after 1 year [17], compared to 39.5% after 1 year in the ARIBON study [18]. The difference in the changes may result from the patients in these two trials having different degrees of bone deficit at the start. Most bone health studies of aromatase inhibitors in breast cancer have followed subjects who had been treated with tamoxifen prior to aromatase inhibitors; however, switching from tamoxifen to an aromatase inhibitor produces a significantly greater increase in bone turnover in these subjects compared with subjects who had no prior tamoxifen treatment [19]. Those studies that examined subjects treated exclusively with aromatase inhibitors reported bone marker data after 1–2 years of treatment in groups of subjects whose mean or median age was 60–69 years [17, 18, 2023]. In these studies treatment with aromatase inhibitors increased bone-resorption markers by 6–40%, with the greatest increase in levels usually seen in the first 6 months of treatment [17, 18, 2023].

Few studies have deliberately investigated older women treated with aromatase inhibitors from the aspect of bone health. Indeed, some studies, such as the ATAC trial bone subprotocol, deliberately excluded women with osteoporosis and were biased against women with osteopenia [17]. Thus, patients over 70 years with breast cancer would be discriminated against because the decline of bone density with age would increase the likelihood of older women having osteoporosis. However, the increasing life span of women following the improvements brought about in breast cancer survival alongside the high incidence of breast cancer in women over 70 years [24, 25] call for studies into the side effects of treatments in older women who may have a significant survival time from breast cancer.

At present, it remains uncertain whether aromatase inhibitors affect bone as much in older patients, in whom age has already had an effect on increasing bone turnover [12, 13], as in younger patients. A study of bone resorption in newly diagnosed breast cancer patients showed that uNTx was low in premenopausal women but not in postmenopausal women [26]. Following treatment with aromatase inhibitors, BMD fell further in early postmenopausal women, <4 years since menopause, than in women further away from menopause [17]. Whether the bone-resorption response to aromatase inhibitors is different in older postmenopausal women vs. younger postmenopausal women is not known. To address this point, we used data collected cross-sectionally to test the hypothesis that the increase in bone-turnover markers in response to anastrozole would be similar in women with breast cancer above and below the age of 70 years. We also measured uNTx levels in a group of women newly diagnosed with breast cancer without evidence of metastatic disease before any procedure was undertaken to treat the cancer in order to ensure that patients with breast cancer do not have markedly different values of uNTx compared with normal subjects.

Materials and Methods

Study Subjects

Female patients with breast cancer about to start or already on treatment with aromatase inhibitors were referred from five local cancer centers to the Charles Salt Centre for Human Metabolism for assessment of their bone status. As well as bone densitometry, uNTx is done routinely on all breast cancer patients since bone-turnover markers may add a dimension to the interpretation of fracture risk as fracture patients have higher levels of bone markers than women without fractures [27, 28]. Patients attending the bone health clinic provide data on menopausal status, prior treatment, and fractures through a self-completed osteoporosis questionnaire. In the present report, postmenopausal women over age 50 years with breast cancer treated with anastrozole (ABC) referred for bone assessment were studied (n = 99). Data were collected between January 2006 and April 2009 on consecutive patients who had been referred by their oncologist either directly or indirectly via a general practitioner. NICE guidelines TA112 (November 2006) state that women with osteoporosis or at risk of osteoporosis should have their BMD formally assessed by bone densitometry at the beginning of anastrozole treatment and at regular intervals thereafter. Those subjects who had received tamoxifen or another aromatase inhibitor therapy prior to initiation of anastrozole or who were currently taking any other treatment known to affect bone metabolism were excluded. Subjects who had taken hormone-replacement therapy (HRT) within the 6 months prior to breast cancer diagnosis were also excluded.

Healthy control (HC) subjects comprised healthy community-dwelling postmenopausal women over the age of 50 (n = 137) recruited through advertisements in local papers for a study of bone turnover. All were healthy, ambulatory, postmenopausal women who were not receiving any treatment known to affect bone metabolism. The study was approved by the local ethics committee, and informed consent was obtained from all subjects.

A study was also undertaken on newly diagnosed breast cancer (NDBC) patients without metastatic disease prior to initiation of any treatment including surgery (n = 88). Consecutive patients were approached at a local breast cancer clinic following diagnosis, and as approved by the local ethics committee, those who consented to taking part sent a urine sample to the laboratory. NDBC women who were already receiving a bisphosphonate were excluded from the study. The local ethics committee approved the studies on NDBC patients without evidence of metastatic disease, and informed consent was obtained. A cross-sectional approach was adopted for the study. Deviations from the age-corrected mean (z score, see below) for the younger and older NDBC patients were compared to those of ABC patients, separated into a group within 6 months of starting anastrozole and a group more than 6 months after starting anastrozole.

Biochemical Measurements

A second-morning void urine specimen was collected by the subject at home prior to the clinic appointment, posted by first-class post to the Charles Salt Centre for Human Metabolism Laboratory on the day of collection, and stored at –20°C upon receipt until analysis, as previously described [29]. Samples were measured routinely as part of a bone marker service once sufficient samples had been received, usually within 2 weeks of receipt. The bone-resorption marker uNTx was measured on a microplate format using an enzyme-linked immunoassay (ELISA) (Osteomark; Ostex International, Seattle, WA) including four quality controls, two provided in the kit and two internal frozen urine quality-control aliquots. These results were corrected for urine volume using creatinine (Cr) levels measured using the Jaffe reaction. Samples were assayed sequentially as part of the bone marker service and validated through the United Kingdom National External Quality Assessment Service (UK NEQAS)-accredited scheme for bone metabolism assays. The UK NEQAS scheme allows the laboratory to assess performance over time and in relation to other laboratories in the United Kingdom also measuring uNTx; our laboratory results lie within ±0.65 standard deviations (SDs) of the target mean. uNTx (nanomoles bone collagen equivalents/millimoles creatinine) levels were log-transformed prior to analysis to ensure a normal distribution: therefore, uNTx is used throughout this report to refer to the log-transformed data. uNTx z scores [(individual result − age-matched HC mean)/age-matched HC SD] were calculated using the log-transformed uNTx levels.

DXA Scanning

BMD was measured on the lumbar spine (LS) and femoral neck (FN) with a QDR 4500 densitometer (Hologic, Bedford, MA). BMD z scores were calculated as above for each site using previously established age-matched levels from healthy postmenopausal women within the local population.

Statistical Analysis

Data were analyzed using SPSS v17 (SPSS, Ltd., Woking, UK). Comparisons between uNTx levels in the subject groups were analyzed using Student’s t-test, while ANOVA was used to investigate differences between age groups. Differences within the ABC group were analyzed by Student’s t-test or the Mann–Whitney U-test if the data were nonparametric or χ2 tests for discontinuous data. Correlations between uNTx and BMD z scores and age and between age and uNTx z score were determined using Pearson’s correlation coefficient (r). Distribution around the age-matched mean of uNTx, LS BMD, or FN BMD was analyzed using the binomial test. The interaction between age and response to anastrozole treatment was compared between women <70 years and >70 years using the method of Matthews and Altman [30]. Statistical significance was taken at the 5% level.

Results

uNTx Levels in Healthy Controls

Values of uNTx by decade were determined in HC subjects (Table 1). In HC subjects uNTx values increased with each decade (P < 0.01), with a positive correlation between age and uNTx (r = 0.321, P < 0.0005). uNTx levels in each decade were used to establish a range from which z scores could be derived.
Table 1

uNTx levels in healthy postmenopausal controls and anastrozole-treated breast cancer patients by decade, excluding women >80 years

Age group

HC

ABC

P

All

1.60 ± 0.24

1.74 ± 0.20

<0.0001

Age (years)

67.4 ± 8.5

67.0 ± 8.0

0.767

 n

137

99

 

50–59 years

1.50 ± 0.25

1.79 ± 0.18

<0.0001

 n

28

21

 

60–69 years

1.58 ± 0.21

1.71 ± 0.24

<0.005

 n

58

37

 

70–79 years

1.69 ± 0.25*,**

1.74 ± 0.18

0.272

 n

40

35

 

ANOVA

   

P

0.0054

0.419

 

Results are mean ± SD. Mean levels of women aged >80 years were not included as the patient numbers were too small (HC = 11, ABC = 6). Statistically significant results are shown in bold

HC healthy controls, ABC anastrozole-treated breast cancer, SD standard deviation

P < 0.005 vs. 50–59 year age group, ** P < 0.005 vs. 60–69 year age group

uNTx Levels in Subjects with Breast Cancer

uNTx levels in NDBC subjects were similar to those in HC subjects up to the age of 69 years (1.56 ± 0.22 vs. 1.55 ± 0.23, nonsignificant [ns]) but significantly lower in subjects over 70 years (1.59 ± 0.14 vs. 1.69 ± 0.24, P < 0.05). There was no difference in uNTx values for NDBC according to age.

ABC subjects had higher values of uNTx compared with HC subjects up to the age of 69 years (P < 0.005 for both ages) (Table 1) but not in the age group over 70 years. ABC subjects also had higher values at all age groups compared with NDBC women (P < 0.05 for all ages).

uNTx z Score in Breast Cancer Patients Treated with Anastrozole

To compare the expected age-related level of uNTx to the level seen in response to anastrozole, we calculated the z score for uNTx levels in the ABC group. In women who had been treated for more than 6 months, by which point other studies have shown the effect of anastrozole on bone to be maximal, the uNTx z score negatively correlated with age (r = −0.473, P < 0.001) (Fig. 1). Examination of the relationship by 10-year age cohorts (50–59 years, etc.) showed that in each of the two earlier decades most ABC patients had a positive uNTx z score (P < 0.01 for each decade), whereas after 70 years uNTx z scores were equally distributed around zero. For subsequent analysis patients were divided into those below and those above 70 years.
https://static-content.springer.com/image/art%3A10.1007%2Fs00223-010-9452-2/MediaObjects/223_2010_9452_Fig1_HTML.gif
Fig. 1

Effect of age on uNTx z score in breast cancer patients treated with anastrozole for more than 6 months. Dotted line indicates the age-matched mean (z score = 0) of healthy control subjects. The uNTx z score was negatively correlated with age in these subjects (r = −0.473, P < 0.001)

uNTx Levels in Patients Above and Below 70 Years with Breast Cancer

The characteristics of ABC patients above and below 70 years are contrasted in Table 2. In the ABC group fewer patients over 70 years had previously received HRT (P < 0.005), chemotherapy (P < 0.0005), or radiotherapy (P < 0.05) than in the younger group. Time since diagnosis, age at menopause, and duration of treatment were similar in the two age groups. The values of uNTx according to age and time since starting anastrozole compared to HC women were significantly higher only in those <70 years (P < 0.0005) (Table 3). In women <70 years anastrozole treatment resulted in a larger difference in uNTx levels compared to HC women of a similar age than in patients >70 years (P < 0.01) (Table 3). This difference in treatment effect was observed in women treated for less than 6 months (P < 0.05) and in those treated for more than 6 months (P < 0.05) (Table 3).
Table 2

Characteristics of anastrozole-treated breast cancer subjects subdivided by age

 

All

<70 years

>70 years

P

n

99

58

41

 

Age (years)

67.0 ± 8.0

61.4 (50.7–68.9)

74.6 (70.2–86.9)

<0.0001

TSD (years)

1 (0–6)

1 (0–6)

1 (0–6)

0.414

Menopausal age (years)

50 (27–57)

50 (27–57)

50 (38–57)

0.831

YSM

17.7 (1.7–39.6)

11.7 (1.7–39.6)

26.7 (13.6–37.2)

<0.0001

Height (m)

1.61 ± 0.05

1.62 ± 0.05

1.58 ± 0.05

<0.0005

Weight (kg)

71.8 ± 14.2

73.3 ± 14.5

69.7 ± 13.7

0.238

BMI

27.8 ± 5.3

27.8 ± 5.1

28.0 ± 5.6

0.843

Proportion of subjects treated with

 Surgery

94 (97.9%)

57 (98.3%)

37 (97.4%)

0.638

 Chemotherapy

25 (26.0%)

23 (39.7%)

2 (5.3%)

<0.0005

 Radiotherapy

53 (55.2%)

38 (65.5%)

15 (39.5%)

<0.05

 Prior HRT

37 (38.1%)

29 (50.9%)

8 (20.0%)

<0.005

 Hysterectomy

27 (27.3%)

17 (29.3%)

8 (24.4%)

0.380

AI treatment time

0.5 (0.08–6)

0.46 (0.08–6)

0.54 (0.08–5)

0.882

Log uNTx

1.73 ± 0.20

1.74 ± 0.22

1.72 ± 0.18

0.645

FN z score

0.05 ± 0.98

0.09 ± 0.98

0.00 ± 0.98

0.675

LS z score

0.36 ± 1.03

0.43 ± 1.09

0.26 ± 0.95

0.421

Results are mean ± SD, median (range), or n (percentage). Differences between <70 and >70 year groups were determined using Student’s t-test or Mann–Whitney U-test if data were nonparametric; statistically significant results are shown in bold

TSD time since diagnosis, YSM years since menopause, BMI body mass index, HRT hormone-replacement therapy, AI aromatase inhibitor, uNTx urinary cross-linked N-telopeptide of type I collagen, FN femoral neck bone mineral density, LS lumbar spine bone mineral density

Table 3

Effects of age and time since starting treatment on changes in uNTx in anastrozole-treated breast cancer patients compared to healthy controls

Group

<70 years

>70 years

ABC

HC

ABC

HC

All subjects

 n

58

86

41

51

 uNTx

1.74 ± 0.029

1.55 ± 0.025

1.72 ± 0.029

1. 69 ± 0.034

 P

<0.0005

ns

 Interaction

<0.01

<6 Months’ treatment with anastrozole

 n

30

86

20

51

 uNTx

1.69 ± 0.037

1.55 ± 0.025

1.67 ± 0.035

1. 69± 0.034

 P

<0.005

ns

 Interaction

<0.05

>6 Months’ treatment with anastrozole

 n

28

86

17

51

 uNTx

1.80 ± 0.037

1.55 ± 0.025

1.78 ± 0.053

1. 69 ± 0.034

 P

<0.0005

ns

 Interaction

<0.05

Results are mean ± SE. Interaction calculated according to the method of Matthews and Altman [29]. Statistically significant results are shown in bold

HC healthy controls, ABC anastrozole-treated breast cancer, uNTx urinary cross-linked N-telopeptide of type I collagen

Figure 2 shows the cross-sectional data for NDBC subjects and for ABC patients treated for <6 months and >6 months according to age. There was no difference in uNTx levels between subjects above and below 70 years in NDBC subjects (1.56 ± 0.22 vs. 1.59 ± 0.14, P = 0.415), ABC subjects treated with anastrozole for less than 6 months (1.69 ± 0.2 vs. 1.70 ± 0.2, P = 0.878), or those treated for more than 6 months (1.79 ± 0.22 vs. 1.75 ± 0.17, P = 0.404). The data do, however, show differences when related to the expected HC values. uNTx was low in NDBC patients over 70 years (see above) but reached the expected age-related values (mean HC >70 years, shown by upper dotted line) after 6 months of anastrozole exposure (Fig. 2). In contrast, patients under 70 years with NDBC exhibited uNTx values consistent with their age, but values were significantly elevated in relation to the expected age-related values (mean HC <70 years, shown by lower dotted line) after 6 months of anastrozole treatment (Fig. 2). The data suggest that the values of uNTx rise equivalently in those below and those above 70 years of age with treatment time.
https://static-content.springer.com/image/art%3A10.1007%2Fs00223-010-9452-2/MediaObjects/223_2010_9452_Fig2_HTML.gif
Fig. 2

Box-and-whisker plot of uNTx levels in newly diagnosed breast cancer patients (NDBC) and breast cancer patients treated with anastrozole for less than and greater than 6 months subdivided according to age. There was no difference between women <70 years (open boxes) and those >70 years (closed boxes) in any of the three groups studied. Dotted lines indicates the age-matched means of healthy control (HC) subjects

BMD z Scores in Patients Treated with Anastrozole

There were no differences in either FN or LS BMD z scores between patients <70 and >70 years (Table 2). However, women in the ABC group were more likely to have a positive LS BMD z score than age-matched controls. The LS BMD z score was above the age-matched mean in 62% of all ABC women (P < 0.05) but was greater (65%) in the <70 year group (P < 0.05) compared with 58% in those aged >70 years. uNTx levels were negatively correlated with both FN BMD z score (r = −0.303, P < 0.005) and LS BMD z score (r = –0.249, P < 0.05).

Discussion

Concordant with earlier reports, uNTx levels in healthy postmenopausal women increase with age [12, 13]. Few reports, however, exist on uNTx levels in postmenopausal women with newly diagnosed breast cancer without metastases who had not received prior therapy including surgery, chemotherapy, or radiotherapy. The ATAC trial included nonrandomized control subjects in the bone subprotocol, but these subjects had undergone primary surgery [17]. A previous report looking at untreated breast cancer subjects found uNTx levels to be similar to healthy postmenopausal controls [27]. In this study we found that uNTx levels tended to be lower in NDBC women >70 years compared to HC women. Values for uNTx might be aberrant; but our uNTx values show long-term consistency with other laboratories within the NEQAS scheme, and the average uNTx levels reported here for healthy normal subjects are similar to other published ranges for Caucasian women [13, 3133]. It is possible that the increased levels of estradiol and free estradiol observed in breast cancer subjects compared to controls [34, 35] may result in lower bone-resorption marker levels in older NDBC subjects.

The increment in uNTx by 16% in the present cross-sectional study of women <70 years on anastrozole for more than 6 months is consistent with increases of 15% and 39.5% reported in other studies on women of a similar age [17, 18]. Similar increments of 16–35% have been recorded in serum CTx [11, 17, 18, 22].

Few studies of aromatase inhibitors in breast cancer have included women over 70 years in any numbers, the majority of studies having a median age <65 years [2023], with one study actively excluding women >75 years [20]. Women over 70 years had less exposure to chemotherapy and radiotherapy, and these treatments could affect uNTx levels; but there were insufficient numbers in the study population to examine this, and most studies have not excluded subjects because of their prior exposure. The finding that uNTx levels in younger women on anastrozole are similar to those in healthy elderly control subjects and to those in elderly women on anastrozole suggests that anastrozole leads to a bone marker profile associated with elderly subjects during treatment.

Although uNTx increased above normal levels with anastrozole therapy in women younger than 70 years, the response was not mirrored in women over 70 years. With increasing age after menopause, total estradiol levels change little but sex hormone globulin levels rise [36, 37], leading to a decline of free estradiol levels with age [38]. As a result, older postmenopausal women have lower levels of free estradiol compared with their younger counterparts. Thus, the effect of aromatase inhibitors on lowering free estradiol levels will, in absolute terms, be less marked than is found in younger women and could lead only to a minimal change in the age-related increase in bone resorption. Unexpectedly, patients with newly diagnosed breast cancer aged >70 years displayed lower uNTx levels compared with control subjects. In one series of patients with breast cancer, bone turnover was inversely related to basal estradiol levels [11]. It is possible that older women with breast cancer have either higher levels of free estradiol or greater sensitivity to the hormone. Our data do not address this possibility, but further work on the relationship between estradiol and bone metabolism in elderly women with breast cancer is merited.

Ultimately, the anxiety engendered through the increase in bone-turnover markers in patients treated with aromatase inhibitors originates from the possibility of increased fracture risk. Heightened bone turnover may be a risk factor for fracture [27, 28], but BMD is also a significant risk factor [39, 40]. Although limited in numbers, our data support the findings that LS BMD is higher than expected in breast cancer [4143], thereby providing some protection against the effects of anastrozole. However, other studies have not found a higher spine BMD in breast cancer patients [44, 45], and there is a similar diversity of findings for the femoral neck [42, 43]. Our study provides some evidence from both the bone marker and the bone density data that women >70 years old taking anastrozole may have a similar risk of fracture as a healthy subject in the general population. Large, long-term studies looking at fracture rates need to be undertaken in order to evaluate this preliminary finding.

This study has several limitations. It is a cross-sectional study that was undertaken to study whether anastrozole affected bone turnover similarly in all postmenopausal age groups. ABC patients were not selected as for a clinical trial, but rather the data were collected as patients were referred to a clinical service for metabolic bone disease. No hormone studies were undertaken, nor was the receptor status of the ABC patients ascertained. We did not exclude subjects who had undergone prior treatment with chemotherapy or radiotherapy. Compliance with and adherence to aromatase inhibitor treatment were not assessed, and we did not account for differences in lifestyle factors due to the limited numbers in this study. The finding that there were differences in bone resorption between the groups may merit further, longitudinal, studies to progress the relationships alluded to in this study.

In conclusion, we cannot reject the null hypothesis that aromatase inhibitors have a similar effect on bone health in younger women, <70 years, than in older women, >70 years. The data suggest that the increase in bone resorption in response to aromatase inhibitors may prematurely age the bone marker profile in younger women. Additionally, they indicate that older women with breast cancer may have lower bone turnover and greater bone density. Whether this relates to estradiol status or receptor sensitivity requires further investigation.

Copyright information

© Springer Science+Business Media, LLC 2010