Background
Breast cancer is the most commonly diagnosed malignancy and the third leading cause of cancer death among women worldwide [
1]. Moreover, despite the application of neoadjuvant and adjuvant therapies to patients in the early stages of breast cancer, this disease remains a major public health challenge. In addition, since bone is the most common site for breast cancer metastasis events [
2], skeletal-related events (SREs) can also develop, and these include hypercalcaemia, bone pain, pathological fractures, and spinal cord compression [
3].
Bisphosphonates (BPs) are potent inhibitors of osteoclast-mediated bone resorption. Consequently, BPs have been used to treat bone metastasis malignancies [
4,
5] and postmenopause osteoporosis as a preventive therapy against SREs [
6]. Furthermore, BPs are increasingly being used for the treatment of early stage breast cancer patients, based on level-one evidence that BPs, particularly zoledronic acid, effectively prevent cancer treatment-induced bone loss (CTIBL) in breast cancer patients receiving chemotherapy and/or estradiol-lowering endocrine therapies [
7‐
10]. Emerging data from preclinical studies have also shown a direct anti-tumour role for zoledronic acid in the inhibition of tumour cell adhesion, invasion, and proliferation, as well as in the induction of apoptosis as demonstrated in multiple human tumour cell lines [
11]. An indirect anti-tumour role for zoledronic acid also includes the inhibition of angiogenesis [
12]. Considerable evidence further suggests that nitrogen-containing BPs exhibit a higher potency by exerting additive or synergistic interactions with standard cytotoxic agents [
13].
Several large-scale randomized controlled trials of adjuvant zoledronic acid therapy for patients with breast cancer have recently been completed, and the results are somewhat controversial [
14‐
17]. Therefore, a comprehensive, systematic review and meta-analysis was performed to evaluate the clinical outcome associated with the use of zoledronic acid as an adjuvant therapy for the treatment of patients with early stage breast cancer.
Discussion
The meta-analysis performed included more than 7,000 patients and provides evidence that the use of zoledronic acid in an adjuvant breast cancer setting improves 5-year OS rates by 14%, reduces the risk of metastasis events by 19%, and lowers the risk of fractures by 25%. In contrast, the use of zoledronic acid did not improve DFS or DMFS, although the 5-year DFS and DMFS rates showed a borderline statistically significant benefit (the P-values were 0.06 and 0.05, respectively).
In a meta-analysis conducted by Zhu et al., oral clodronate was found to increase 5-year OS rates and BMFS rates [
24], and our results are consistent with these outcomes. A possible explanation for the beneficial role of BPs is the direct and indirect anti-tumour activities of these compounds, which include the prevention of tumour cell adhesion to bone, the induction of tumour cell apoptosis, and an inhibition of angiogenesis [
11,
12]. In addition, some clinical trials have shown that zoledronic acid is able to eliminate disseminated tumour cells (DTCs) from bone marrow [
25,
26], thereby reducing the tumour burden in bone and preventing the development and progression of bone metastases. Zoledronic acid, as a third-generation BP, is characterized by an imidazole ring containing two nitrogen atoms, and appears to mediate stronger anti-tumour effects compared with non-nitrogen-containing BPs. For example, the presence of nitrogen has been associated with synergistic interactions with standard cytotoxic agents [
13] and the regulation of T-cell-mediated host immunosurveillance of tumour cells [
27]. There is also evidence that zoledronic acid may inhibit visceral metastases [
28]. The Southwest Oncology Group (SWOG)-S0307 trial [
29] is ongoing to compare clodronate, zoledronic acid, and ibandronate therapies in 4,500 patients with stage I–IIIa breast cancer. The goal is to establish the best choice of BPs in an adjuvant setting. In addition, the SUCCESS study [
30] involving 3,700 patients in the early stages of breast cancer has been designed to compare DFS for patients undergoing chemotherapy and receiving zoledronate (4 mg) for 2 years versus 5 years. The optimal treatment time and the largest cost-benefit ratio for this cohort are also being examined. Another type of antiresorptive agent that is being evaluated is denosumab, a fully humanised monoclonal antibody that has been associated with a metastasis-prevention role and is currently being investigated in adjuvant phase III trials, including ABSCG-18 (ClinicalTrials.gov Identifier, NCT00556374) and D-CARE (ClinicalTrials.gov Identifier, NCT01077154).
In the current meta-analysis, zoledronic acid use was not found to benefit DFS and DMFS. However, the largest studies performed to date [
14‐
16,
31] have shown statistically significant benefits for the DFS and OS of postmenopausal women. Similarly, the AZURE trial [
14] included a prespecified subgroup analysis based on age and menopausal status. For patients who had undergone menopause for more than 5 years before their breast cancer diagnosis, a significant improvement in DFS and OS was observed when zoledronic acid was received. In contrast, no benefit was observed for premenopausal and perimenopausal patients, or patients with unknown menopausal status, who also received zoledronic acid. In the ABCSG-12 trial [
15], significant survival benefits were observed for patients > 40-years-old with low estrogen levels resulting from gonadotropin-releasing hormone use. In the final 60-month follow-up of the ZO-FAST trial [
16], a statistically significant benefit for DFS (hazard ratio (HR) = 0.63;
p-value = 0.0516) and OS (HR = 0.50;
p-value = 0.0224) was associated with zoledronic acid use for patients who were postmenopausal for more than five years or who were > 60 years of age when starting the study. Furthermore, based on an analysis of stratification subgroups by the National Surgical Adjuvant Breast and Bowel Project protocol (B-34) [
31], OS, BMFS, and non-BMFS were found to be significantly improved with clodronate treatment in women aged 50 years and older. In comparison, there was little difference observed between the clodronate group and placebo group, each containing patients younger than 50 years. In a meta-analysis conducted by Yan and Yin [
32], a better DFS outcome (RR = 0.763;
p-value < 0.001) was observed, yet a reduced risk of distant (RR = 0.744;
p-value = 0.003) and locoregional recurrence (RR = 0.508;
p-value = 0.001) was found for the postmenopause subgroup with zoledronic acid adjuvant use. These findings suggest that the anticancer benefits of BPs, especially zoledronic acid, have the greatest potential in a low-estrogen environment. While the underlying mechanism for this observation remains unclear, it has been hypothesized that estrogen’s effects on the bone microenvironment play a key role in determining who benefits from adjuvant BP therapy.
The results of the present study are consistent with those of the meta-analysis conducted by Valachis et al. [
33]. However, there are additional considerations regarding this comparison. First, the ZO-FAST study included in the current meta-analysis included data from a 60-month follow-up period, while the Valachis study analyzed data from a 54-month follow-up period. Secondly, the ABCSG-12 trial included 62- and 84-month follow-up data, while 62-month follow-up data were used in the present study. As a result, a more reliable evaluation of 5-year OS outcome was achieved. Third, while Valachis et al. analysed the OS for only five studies, and these did not include the Z-FAST and E-ZO-FAST trials which had negative results, the present meta-analysis did include all of these studies. Correspondingly, the present study included a robust sensitivity analysis. However, our review had several limitations to consider as well. First, our analysis was based on published or presented data, and did not include individual patient data. Therefore, certain details and prespecified subgroup data could not be acquired. For example, the AZURE trial [
14] did not provide distant metastases data, thereby resulting in an incomplete analysis of DMFS (p-value = 0.05) for 3,995 patients. Secondly, subgroup analysis was not performed according to menopausal status since the definition of postmenopausal status was found to vary in the articles examined. Furthermore, the majority of the subgroup analyses performed were unplanned, making comparisons unreliable. Thirdly, some of the trials considered [
16,
17,
22,
23] included a control group that received a delayed administration of zoledronic acid. Moreover, if this treatment was administered in the lumbar spine (LS) or total hip (TH), the associated T score fell below -2.0, or a non-traumatic fracture occurred. Therefore, a number of patients in the control group received zoledronic acid during follow-up, and it cannot be excluded that the effect of zoledronic acid on patient outcome was underestimated. In addition, these trials were designed to assess changes in bone mineral density (BMD) as a primary endpoint, and not DFS or OS. Fourthly, the clinical trials included in this meta-analysis utilized different methodologies, did not conduct double-blinded studies, and the schedule and duration of zoledronic acid use differed. Finally, sensitivity analysis was performed by omitting two trials having 36-month follow-up periods [
22,
23], and this omission affected the results obtained. It is hypothesized that the reduced number of event outcomes associated with this shorter follow-up period lacked statistical efficiency. Furthermore, the studies in the meta-analysis were found to have an unclear risk of bias or a high risk of bias, and the latter may have inflated the results obtained.
Materials and methods
Data sources
A systematic search of the PubMed and EMBASE databases was performed of entries made through 12 July 2013. Online abstracts from the proceedings of the Annual Meetings of the American Society of Clinical Oncology (ASCO) (1992–2013) and the San Antonio Breast Cancer Symposium (SABCS) (2004–2013) were also reviewed. The terms, 'bisphosphonates’ , 'zoledronic acid’ , 'adjuvant’ , and 'breast cancer’ were used as key words.
Study selection
A study was considered acceptable according to the following inclusion criteria: (a) it represented a prospective randomized trial, (b) the patients examined were in the early stages of breast cancer (e.g., stage I-III), (c) zoledronic acid was used in an adjuvant setting, (d) sufficient data were reported to allow an estimate of RR to be calculated for DFS and OS, and (e) English was the language of the publication. When multiple articles regarding the same trial were retrieved, the most recent or most complete publications were selected. Reports involving advanced breast cancer (e.g., recurrent breast cancer or metastatic breast cancer) were excluded. Non-randomized studies were also excluded, as well as letters to the editor, reviews, abstracts containing insufficient detail to meet the inclusion criteria, articles published in a book, and papers published in a language other than English.
Clinical endpoints
Overall, eight studies including 3866 subjects and 3864 controls met our search criteria and were evaluated. For each of these studies, the first authors’ name, year of publication, number of patients randomly assigned, patient menopausal status, mean patient age, medication strategies for treatment and control arms, and possible outcomes at final follow-up were recorded. Primary endpoints for these studies included OS and DFS, while secondary endpoints included BMFS, DMFS, and FFR. Furthermore, the outcome measures were based on an intention-to-treat analysis. The methodological qualities of each study were evaluated using modified Jadad scoring (ranging from 0 to 7), with 0–3 indicating low quality and 4–7 indicating high quality. Cochrane Collaboration’s tool for assessing risk of bias was also employed, which includes low, unclear, and high categories of risk of bias. Any differences in assessments made by M.F.H and W.D.F were resolved through discussion.
Statistical analysis
The Mantel-Haenszel method was used to calculate RRs and their 95% CIs using either fixed- or random-effects models, depending on the amount of heterogeneity observed. The effect of heterogeneity was quantified using
I2 = 100% × (Q–df) / Q, where Q is Cochran’s χ
2 heterogeneity statistic, df is the degrees of freedom, and
I2 ranges between 0% and 100%. Moreover,
I2 values of 25%, 50%, and 75% are classified as low, moderate, and high estimates, respectively [
34]. Statistically significant heterogeneity was defined according to chi
2, with a
P value < 0.1 or an
I2 value > 50%. A random-effects model meta-analysis for heterogeneous outcomes and a fixed-effect model meta-analysis for homogeneous outcomes were also performed [
35]. For the sensitivity analysis performed, two studies were excluded that had follow-up periods less than five years.
Funnel plots were generated to investigate potential publication bias. All p- values were two-sided and a RR < 1 favored the use of zoledronic acid. All analyses were performed using Review Manager (Version 5.2. Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2012).
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
MFH contributed to the design of this study, literature review, data collection and analysis, and drafting of this manuscript. WDF contributed to the literature review, data collection and analysis, and critical revisions of the manuscript. XQZ contributed to the conceptualization of this study, data interpretation, and provided scientific advice. All authors have read and approved the final manuscript.