Letrozole is safe and effective in postmenopausal women with estrogen receptor-positive early breast cancer, but long-term aromatase inhibitor use may cause bone loss and increase fracture risk. This study evaluated an immediate or delayed strategy of bone protection therapy with zoledronic acid.
A total of 1065 patients who were receiving adjuvant letrozole were randomized to immediate-start or delayed-start zoledronic acid (4 mg intravenously biannually for 5 years). The delayed group received zoledronic acid if lumbar spine or total hip T-score decreased below −2.0 or when a nontraumatic fracture occurred. The primary endpoint was change in lumbar spine bone mineral density (BMD) at Month 12. Secondary endpoints included changes in total hip BMD, serum bone turnover markers, and safety at Month 12.
Lumbar spine BMD increased from baseline in the immediate arm, while it decreased from baseline in delayed-arm patients. At Month 12, the differences between the groups in lumbar spine and total hip BMD were 5.7% (P < .0001; 95% confidence intervals [CI], 5.2% to 6.1%), and 3.6% (P < .0001; 95% CI, 3.3 to 4.0%), respectively. Both regimens were well tolerated with few serious adverse events. Bone pain was higher in the immediate group, as expected, because some patients experienced acute-phase reactions after zoledronic acid infusion.
Estrogen is essential for maintenance of bone mass in adult women,1 and the decrease in estrogen levels at menopause results in increased bone turnover and loss.2 Early menopause, induced by either chemotherapy or ovarian suppression, is associated with premature bone loss.2 In postmenopausal women with breast cancer, aromatase inhibitors (AIs) result in a nearly complete suppression of aromatase activity and a fall in circulating estrogen concentrations to undetectable levels.3, 4 In the adjuvant setting, all third-generation AIs are associated with further acceleration of bone loss and an increased rate of bone fracture, compared with tamoxifen or placebo.5–8 Efforts to avoid fractures are needed. In the Arimidex, Tamoxifen, Alone or in Combination (ATAC) trial, the fracture risk increased during therapy but returned to normal after stopping anastrozole.9 Maximal bone mineral density (BMD) loss occurs during the first year after treatment-induced menopause or initiation of an AI, making the concept of bone preservation during this time an important area of study.10
Fracture risk is evaluated in this study through the use of a surrogate endpoint, change in BMD. Several studies have shown that BMD is a strong surrogate marker for fracture risk in postmenopausal women.11, 12 A meta-analysis of approximately 39,000 men and women participating in 12 osteoporosis trials reported a 1.6-fold to 2.2-fold increase in fracture risk for each standard deviation (SD) fall in BMD T-score for women.13 (A T-score uses the young-normal standard of optimal or peak density of a 30-year old healthy adult rather than age-matched comparisons of BMD).
The use of an AI, either alone or sequentially after tamoxifen, is emerging as the optimal adjuvant endocrine therapy for postmenopausal women with hormone receptor-positive breast cancer. Compared with tamoxifen alone, AIs have improved disease-free survival and produced fewer gynecological or thromboembolic adverse events. Strategies to prevent bone loss, particularly in women using such therapy after chemotherapy or a premature menopause, are needed. The Austrian Breast and Colorectal Cancer Study Group trial 12 (ABCSG-12) reported a 17.3% loss in BMD in premenopausal patients with early breast cancer who were receiving anastrozole and a luteinizing hormone-releasing hormone (LHRH) agonist after 36 months of treatment.10 Zoledronic acid, a nitrogen-containing bisphosphonate, inhibits osteoclast-mediated bone resorption and maintains or increases BMD.14 Recent data have also indicated that women with established postmenopausal osteoporosis, receiving an annual 5 mg dose of zoledronic acid, experienced a 70% risk reduction in new spine fractures (P < .0001) and a 40% risk reduction in hip fractures (P = .0032).15
The aim of the ZO-FAST trial was to evaluate the effect of either immediate or delayed use of 4 mg zoledronic acid on the prevention of bone loss in both recently postmenopausal (because of chemotherapy or LHRH suppression) and established postmenopausal women with early stage breast cancer who had received adjuvant letrozole for 5 years.
MATERIALS AND METHODS
Postmenopausal women from 112 centers in 28 countries who had estrogen receptor-positive early breast cancer and baseline lumbar spine (LS) and total hip (TH) T-scores above −2.0 were included. Eligibility criteria was similar to the Z-FAST study.16
In this open-label, multicenter, randomized study, patients orally received 2.5 mg letrozole daily for 5 years or until disease progression and were randomized to intravenously receive immediate or delayed 4 mg zoledronic acid for 15 minutes every 6 months for 5 years. The immediate group received zoledronic acid after randomization, whereas the delayed group received zoledronic acid when 1) postbaseline spine or hip T-score decreased to below −2.0; 2) a nontraumatic clinical fracture occurred; or 3) asymptomatic fracture was discovered at the Month-36 scheduled visit. Patients were instructed to take an oral calcium supplement (500 mg) and a multivitamin tablet containing vitamin D (400 IU to 800 IU) daily during the study. Patients were stratified according to adjuvant chemotherapy (yes vs no), baseline T-score (normal, T-score above −1.0, or T-score between −1.0 and −2.0), and menopausal status (recently postmenopausal vs established postmenopausal).
The primary endpoint of this study was the percentage change in spine BMD at 12 months in patients who were receiving immediate-start compared with delayed-start zoledronic acid. The secondary endpoints included the percentage change in total hip (TH) BMD, changes in serum N-telopeptide (NTX), and bone-specific alkaline phosphatase (BSAP) concentrations at 12 months. BMDs of the spine and hip were evaluated at baseline, 6 months (optional), and 12 months, and will be evaluated at 24, 36, and 48 months, and at the final visit by using either Hologic (Hologic, Bedford, Mass), Lunar (GE Medical Systems Lunar Corporation, Madison, Wis), or Norland (Norland, Fort Atkinson, Wis) dual-energy x-ray absorptiometry (DXA) devices. T-scores were calculated by using manufacturer-specific T-score databases. All DXA devices were standardized and cross-calibrated by using 4 Bio-Imaging Bona Fide Phantoms (CIRS Tissue Simulation & Phantom Technology, Norfolk, Va). Enrollment eligibility and timing for initiation of zoledronic acid in the delayed group were based on local DXA readings; however, a central reader (BioImaging, Leiden, Netherlands) analyzed all DXA scans for the primary analysis. Patients with baseline and Month-12 spine BMD measurements were included in the BMD efficacy analysis. BMD data were not available for all patients for the Central Reader. This was primarily due to technical reasons at both the site and at the Central Reader. Also, some patients were not compliant. This affected both baseline and Month-12 assessments. These problems caused some of these patients not to be included in the analysis of BMD data. However, these women were included in the adverse event summary.
Serum NTX concentrations, measured by using the Osteomark and Wampole NTX assays (Wampole Laboratories, Princeton, NJ), and BSAP concentrations, measured by using the Metra immunoassay (Quidel Corporation, San Diego, Calif), were evaluated in a subset of patients by a central laboratory (CRLMedinet Laboratory, Inc., Breda, Netherlands) at baseline and every 3 months during the first year and will be measured every 6 months during years 2 to 3, once at 48 months, and at the final visit. The samples were obtained in the morning while patients were fasting. The normal range for NTX is 6.2 nmol BCE/L to 19.0 nmol BCE/L for all age groups, including postmenopausal women. The BSAP normal range for postmenopausal women is 14.2 U/L to 42.7 U/L. Adverse events and disease progression were evaluated every 6 months. Adverse events were graded by using the National Cancer Institute Common Toxicity Criteria (NCI-CTC), version 3.0.17 Serum creatinine levels were measured at baseline, every 6 months, and at the final visit. Lipid levels were collected on a subset of patients.
The ethics committee at each participating institution approved this study. Informed consent was obtained from each patient before enrollment. The study was funded by Novartis Oncology (East Hanover, NJ). The steering committee directed the 12-month data review and analysis; all data analyses were performed by PRA International, Mannheim, Germany.
The study design used a 2-sample Student t test, with a power of 90% and a significance level of P = .05, to detect a 4% difference in percentage change in spine BMD with a standard deviation of 18% from baseline to 12 months between the groups. A sample size of 427 patients per treatment arm was required. To allow for a 5% dropout rate, at least 450 patients in each treatment arm were required.
Unless otherwise specified, all statistical tests were performed by using a significance level of P = .05 against a 2-sided alternative hypothesis. Analyses of BMD and bone markers were based on the intent-to-treat (ITT) population, which comprised all randomized patients. Analyses related to cancer recurrence used a modified ITT (mITT) population, defined as all randomized patients who underwent at least 1 postbaseline assessment. The safety population included all patients who received at least 1 dose of zoledronic acid or letrozole. Analyses of adverse events and clinical fractures used the safety population according to randomized treatment. More detailed analyses (including subset analyses) related to bone health, including BMD, NTX, and BSAP, used the safety population according to treatment received. Patients who discontinued the study early were instructed to return for a final assessment of BMD, serum NTX, and BSAP concentrations, and adverse events 4 weeks after treatment cessation.
The primary analysis was performed after all patients had passed the 12-month visit. An analysis of covariance (ANCOVA) model, with adjuvant chemotherapy use and region as factors and baseline BMD and years since menopause as covariates, was used to compare differences between groups; Student t tests for paired data were used to compare differences within treatment groups in spine and hip BMD and serum NTX and BSAP concentrations from baseline to Month 12.
The study was not designed or powered to detect a difference in the incidence of clinical fractures or breast cancer relapse. The frequency of adverse events is reported for both groups.
Between May 2003 and August 2004, 1065 patients were randomized by 112 centers in 28 countries to receive either immediate or delayed zoledronic acid (Fig. 1). The baseline characteristics of the 2 groups were similar (Table 1).
Table 1. Patient Demographics
A T-score uses the young normal standard of optimal or peak density of a 30-year old healthy adult rather than age-matched comparisons of BMD.
ECOG indicates Eastern Cooperative Oncology Group; SD, standard deviation.
Based on local dual-energy x-ray absorptiometry scan results.
Patients randomized; ITT population
Patients in modified intent-to-treat; mITT population
Patients in safety population; randomized treatment
Screening failures were collected as a part of the automated patient randomization system and 446 screens were reported. The reasons were unacceptable test procedure result, including low-baseline BMD score, 289 (65%); other, 50 (11%); patient withdrew consent, 37 (8%); patient did not meet the diagnostic and/or severity criteria, 22 (5%); unacceptable laboratory value, 22 (5%); unacceptable past medical history and/or concomitant diagnosis, 16 (4%); unacceptable use of excluded medication and/or therapies, 4 (1%); intercurrent medical event, 4 (1%); and unknown, 2 (0.5%).
Of the 1065 randomized patients, 1 patient did not undergo the postbaseline assessment because she withdrew consent. The demographics of the patient population are shown in Table 1.
In the delayed arm, 75 patients had started zoledronic acid by the time of the 12-month analysis. Of these, 54 were started based on the lumbar or hip BMD T-score falling to less than −2, and 1 patient had experienced a nontraumatic clinical fracture. Twenty patients were started because of misunderstanding of the protocol at the site level.
Bone Mineral Density
For lumbar spine (L1-L4) BMD, in the ITT population, the unadjusted mean percentage change from baseline in the immediate arm was 2.1%, and in the delayed arm, it was −3.5%. Both of these mean changes were significantly different from zero (P < .0001). At Month 12, the adjusted (least-squares) mean percentage difference in BMD between the 2 groups was 5.7% at the spine (P < .0001; 95% confidence interval [CI], 5.2–6.1) and 3.6% at the hip (P < .0001; 95% CI, 3.3–4.0), respectively (Fig. 2). Similar results were seen for the safety population, for lumbar spine (L2-L4) with an adjusted mean percentage change from baseline in the immediate arm of 1.9% (95% CI, 1.6–2.2) compared with −3.1% (95% CI, −3.5 to −2.8) in the delayed arm (P < .0001).
In the recently postmenopausal women included in the safety population, immediate zoledronic acid prevented BMD loss at both L2-L4 lumbar spine and TH with an adjusted mean percentage change of −0.2% (95% CI, −1.1 to 0.7) and 0.07% (95% CI, −0.6 to 0.7)at 12 months, respectively, whereas in those randomized to delayed intervention, decreases in BMD of 5.5% (95% CI, −6.3 to −4.7) and 3.4% (95% CI, −4.0 to −2.8) at the lumbar spine and TH at 12 months, respectively, were seen (P < .0001) (Fig. 3). These results show a trend toward more rapid bone loss in the recently postmenopausal women versus established postmenopausal women. This is demonstrated by looking at the delayed arm in the 2 groups. The losses in BMD are −5.49 in the recently postmenopausal group versus −2.61 in the established postmenopausal group.
At baseline, 737 (69.2%) patients had normal spine BMD (370 patients, 69.5% in the immediate group; 367 patients, 68.9% in the delayed group) (Table 1). After 12 months, a higher percentage of patients in the delayed group with normal baseline BMD developed mild to moderate osteopenia (T-score lower than −1 but higher than −2) compared with patients in the immediate group (19.1% vs 0.9%) (Table 2). The difference in distributions was statistically significant (P < .0001). At baseline, 162 (30.5%) patients in the immediate group and 166 (31.1%) patients in the delayed group already had mild to moderate osteopenia at the spine and/or hip, placing them at a higher risk for progressing to severe osteopenia (T-score lower than −2.0 but higher than or equal to −2.5) or osteoporosis (T-score lower than −2.5) with letrozole therapy (Table 1). In the delayed group, a higher percentage of patients progressed from mild or moderate to severe osteopenia compared with the immediate group (18.0% vs 0.6%) (Table 2). The difference in distributions was statistically significant (P < .0001).
Table 2. Shift Table for Lumbar Spine BMD Measurements at Month 12*
A T-score uses the young-normal standard of optimal or peak density of a 30-year old healthy adult rather than age-matched comparisons of BMD.
BMD indicates bone mineral density.
Based on central reading of dual-energy x-ray absorptiometry scan results.
Baseline BMD data for 17 patients in the immediate group and 24 patients in the delayed group were not transferred to or could not be read by central reader from study sites; therefore, they not included in the analysis.
n = 342
n = 341
Normal (T-score >−1)
Mild osteopenia (T-score ≤−1 to ≥−2)
Severe osteopenia (T-score <−2)
n = 162
n = 150
Normal (T-score >−1)
Mild osteopenia (T-score ≤−1 to ≥−2)
Severe osteopenia (T-score <−2)
n = 11
n = 17
Normal (T-score >−1)
Mild osteopenia (T-score ≤−1 to ≥−2)
Severe osteopenia (T-score <−2)
During the first year, the incidence rate of fractures was similar, with 8 (1.5%) patients in the immediate group and 9 (1.7%) patients in the delayed group.
Breast Cancer Recurrence
Twelve (2.3%) patients suffered distant recurrences in the immediate group as did 11 (2.1%) patients in the delayed group. There were 4 deaths in the immediate arm and 5 deaths in the delayed arm due to breast cancer relapse.
Markers of Bone Turnover
Serum NTX and BSAP concentrations were measured in a subset of 293 patients with baseline characteristics similar to the entire study population. By Month 12, the unadjusted mean of percentage change from baseline in serum BSAP concentrations had decreased significantly (P < .0001) in the immediate group (−15.5%) and increased significantly (P < .0001) in the delayed group (30.1%). The difference between the immediate and delayed groups in unadjusted mean percentage change from baseline of the bone turnover markers at Month 12 was 45.6% (P < .0001) for serum BSAP (Fig. 4).
The prespecified endpoint for NTX results was the percentage change from baseline at Month 12. The assay used to determine NTX results changed during the course of the study. The 2 assays were not standardized against each other. Therefore, the ability to interpret the magnitude of the change from baseline is limited. Nevertheless, given that treatments were randomized, the magnitude of the treatment difference on absolute NTX scores may still be meaningful, and the P value for the significance of the treatment difference is meaningful. In fact, the use of 2 different assays would have the effect of increasing variability of the treatment difference, which would tend to reduce statistical significance. Then, a statistically significant difference would be a robust finding. The unadjusted mean (SD) NTX value at Month 12 was 22.8 (8.02) nmol BCE/L for the immediate group and 30.3 (9.50) nmol BCE/L for the delayed group, for a mean treatment difference of 7.5. As a result, the NTX concentration is, on average, 33% higher in the delayed arm than in the immediate arm. The value for the treatment difference, comparing the 2 groups on computed percentage change from baseline is P < .0001.
Adverse events having an incidence of at least 5% in either treatment group are shown in Table 3. The occurrence of adverse events was similar between the groups with the exception of bone pain, which was higher in the immediate zoledronic acid group (12.3% vs 6.9%) than the delayed group, as expected, because of the acute-phase reaction that is known to occur in some patients after zoledronic acid infusion. Only 2 patients experienced renal impairment, and both cases were unrelated to study medication as neither had started zoledronic acid at the time of the event.
Table 3. Adverse Events Occurring in Greater Than 5% of Patients
No. of patients (%)
Immediate group n = 527
Delayed group n = 525
Pain in extremity
Serious adverse events occurred in 8.2% and 6.7% of patients in the immediate and delayed groups, respectively. Withdrawal from the study due to adverse events occurred in 5.3% of patients in the immediate group and 4.7% in the delayed group; 1.5% and 1.0% of patients in the immediate and delayed groups, respectively, discontinued therapy because of serious adverse events.
The mechanism between osteonecrosis of the jaw and treatment regimens including bisphosphonates has not been established. Although, osteonecrosis of the jaw was not specifically defined in the protocol, the study database was searched for adverse events that could capture this type of event. The search identified 1 patient. This patient who had been randomized to the immediate arm had experienced a delay in dental healing (grade 2) of 60 days duration. The event was suspected to be related to the study medication; however, no action was taken, and the event resolved without study drug interruption.
Adjuvant AI therapy administered either alone for 5 years, or sequentially after 2 to 5 years of adjuvant tamoxifen, improves disease-free survival in postmenopausal women with endocrine-responsive breast cancer when compared with tamoxifen alone for 5 years.5–8, 18 In general, AIs are associated with a favorable safety profile with fewer thromboembolic and gynecological adverse events than are seen with tamoxifen. However, uniformly reported disadvantages of AIs compared with tamoxifen therapy are the development of accelerated bone loss (decline in BMD) and increased fracture risk.19
Bisphosphonates are an attractive solution for the prevention of cancer treatment-associated bone loss in postmenopausal women with early breast cancer. Zoledronic acid is the only compound that has been extensively studied and has proven efficacy in the context of AI-associated bone loss.16 The rates of bone loss in the presence of an AI are approximately 5-fold higher than those observed in postmenopausal women with established osteoporosis in the absence of AI therapy. This difference led to the dose selection of 4 mg every 6 months in cancer patients.20
The fracture efficacy of zoledronic acid has been demonstrated in women with established postmenopausal osteoporosis, who received an annual 5 mg infusion of zoledronic acid.15 Therefore, it is reasonable to assume that there would be fracture efficacy in the population with AI-associated bone loss treated with zoledronic acid on a slightly more frequent schedule of 4 mg every 6 months.
Recently, the first results from the Z-FAST study demonstrated that patients who received immediate zoledronic acid every 6 months were less likely to develop clinically significant bone loss at 1 year than women who received delayed. zoledronic acid.16 We have confirmed these results in the ZO-FAST study with an increase in BMD in the immediate zoledronic acid group compared with a significant reduction in BMD in the delayed zoledronic acid group. Recently postmenopausal patients with chemotherapy or LHRH analog-induced menopause have a higher rate of bone loss when combined with AI therapy. This was demonstrated on the delayed arm. In recently postmenopausal women had a BMD decline of −5.49 versus −2.61 in the established postmenopausal women. The immediate strategy of zoledronic acid 4 mg every 6 months was able to maintain BMD at both the spine and the hip (Fig. 3).
The ABCSG-12 trial10 studied premenopausal patients who were undergoing treatment with an LHRH analog (goserelin) and randomized to either tamoxifen or the AI, anastrozole, with or without zoledronic acid, to assess the long-term bone safety of the combined hormonal treatment. BMD change was used to define the efficacy of zoledronic acid. This study confirmed that early combined hormonal therapy-induced menopause resulted in very substantial bone loss, mainly evident during the first 12 months of treatment. The magnitude of BMD loss was more pronounced with the combination of goserelin + anastrozole than with goserelin + tamoxifen but was effectively prevented by immediate administration of zoledronic acid.10
The ZO-FAST study shows that there are postmenopausal women who are at significant risk of developing severe osteopenia within the first year of AI therapy in the delayed zoledronic acid treatment group. All of these patients with a mild degree of osteopenia at the commencement of letrozole are less able to compensate for the increased bone turnover and consequent bone loss associated with AI therapy. By Month 12, 10% of patients in the delayed group required commencement of zoledronic acid because of a fall in BMD with a T-score less than −2.0.
Only 17 fractures were noted in this analysis, 8 in the immediate arm and 9 in the delayed arm. We did not expect to see a significant difference in fractures at this time because of the short follow-up and the number of patients enrolled. Analysis of fractures and breast cancer recurrence data will be closely monitored for 5 years.
This trial defined the cutoff point for entry BMD to a T-score of greater than −2. Sixty-five percent of screening failures could not be enrolled for this reason. Therefore, patients with severe osteopenia or osteoporosis at the initiation of adjuvant AI therapy represent a significant patient population. The optimal way to maintain bone health in such patients remains an important clinical issue.
The changes in bone marker measurements observed in a subset of patients who were receiving immediate versus delayed zoledronic acid suggest that the effect of zoledronic acid on bone turnover is both rapid and sustained over at least the first 12 months. The mean of percentage change from baseline in BSAP significantly decreased in the immediate zoledronic acid treatment group but significantly increased in the delayed treatment group. In our study, the unadjusted mean difference in BSAP concentrations between the immediate and delayed treatment groups was 8.9 U/L at Month 12. From a bone safety viewpoint, although BSAP was suppressed in the immediate group, the mean concentrations at each time point remained within the normal premenopausal reference range, which is consistent with continued and active bone remodeling (Fig. 4).
Zoledronic acid was generally well tolerated, with few discontinuations in either group. Bone pain, pyrexia, and headache, which are consistent with the acute-phase reaction that is known to occur in about 30% of the patients who receive zoledronic acid infusion, were more common in the immediate than in the delayed group, but the intensity of such adverse events was mild to moderate. Only 2 patients experienced renal impairment, which was not related to study medication. The reported occurrence of osteonecrosis of the jaw in women with advanced breast cancer receiving chronic monthly intravenous bisphosphonates, among other therapies, is an infrequent event. The incidence of this complication has not been determined prospectively, but current estimates suggest an overall frequency of about 1%.21, 22 To date, no suspected cases of osteonecrosis of the jaw were reported in this study.
National Osteoporosis Foundation (NOF), World Health Organization (WHO), and the American Society of Clinical Oncology (ASCO) have published guidelines for the management of osteoporosis.2, 23, 24 The WHO guidelines recommend treatment for patients when their T-score is less than or equal to −2.5 or for osteopenic patients with additional strong risk factors.2 NOF recommends treating patients when their T-score is less than −2.0 with no risk factors or if the T-score is greater than or equal to −1.5 with 1 or more risk factors or a prior vertebral or hip fracture.23
ASCO recommends that all women at high risk of osteoporosis, including women who are receiving AI therapy, undergo baseline assessment of BMD and regular follow-up while on treatment.24
The 12-month results of this ongoing clinical trial suggest that initiation of intravenous 4 mg zoledronic acid every 6 months in postmenopausal breast cancer patients receiving adjuvant AI therapy is an effective treatment to prevent bone loss. It is particularly appropriate for those women with osteopenia at baseline because they have a significant risk of developing osteoporosis during treatment with an AI. Women recommended for AI therapy who have additional risk factors for fracture, such as advanced age, prior history of fracture as an adult, and personal or family history of osteoporosis, may be best managed by an immediate rather than a delayed bone protection strategy. For women with normal BMD at baseline scheduled to receive an AI, the risk of developing osteoporosis appears to be low. No patients in the ATAC study or Intergroup Exemestane Study bone subprotocols who commenced AI therapy with normal BMD became osteoporotic while on treatment.25, 26 In the ZO-FAST study, follow-up is short, but the risk of transition from normal to severe osteopenia also appears to be very low. At the time of this primary analysis (12 months), no patients with normal BMD at baseline had developed severe osteopenia, even in the absence of bone protection with zoledronic acid.
Longer follow-up is needed to determine whether the bone loss observed in the delayed group can be stabilized or restored to baseline values with the subsequent administration of zoledronic acid. The 2-year, 3-year, and 5-year results of this trial and other ongoing clinical trials are necessary to further assess the impact on BMD loss and fracture rates in patients with early stage breast cancer who are receiving long-term adjuvant AIs.