Oral Bisphosphonates and Improved Survival of Breast Cancer

Bisphosphonates are indicated for treatment and control of osteoporosis by inhibition of osteoclast-mediated bone resorption and reduction in calcium release into the blood stream (1) and are being used by millions in the world, most of them postmenopausal women (2). Third-generation nitrogen-containing bisphosphonates are now proven to effectively treat bone metastases in women with breast cancer (3). It is also hypothesized that bisphosphonates can prevent bone metastases from occurring in women with breast cancer if given in an adjuvant setting, with evidence that is currently, after long debates, evaluated as consistent (4–12). Sustained use of oral bisphosphonates is also possibly associated with reduced risk of breast and colorectal cancers. Three case–control studies found a strong negative association between the use of common second-generation oral bisphosphonates prior to diagnosis and the risk of breast cancer (13–15) with evidence in another study that these agents also reduce the risk of contralateral breast cancer (16). However, a recent publication based on secondary analysis of data from historical randomized trials did not find a similar association (17). In addition, 3 different studies found that postmenopausal bisphosphonate users had a markedly reduced risk of developing colorectal cancer (18–20). Recently, zoledronic acid was found to improve disease-free survival and overall survival in several trials of adjuvant treatments for breast cancer (AZURE, ABCSG-12, ZO-FAST) including in premenopausal women and women with artificially induced postmenopause (21–23) and prolong survival in patients with multiple myeloma and other advanced cancers (24–27).

A variety of direct and indirect putative anticancer activities of bisphosphonates have been studied in animal models. Such anticancer activity has been postulated to involve cancer cells in the bone microenvironment (28–30) or be more generalized and not restricted to the bone microenvironment or to the antiresorptive qualities of the bisphosphonates (27, 31–37). Nitrogen-containing bisphosphonates inhibit protein prenylation through blocking of the mevalonate pathway. Isoprenoid biosynthesis is required for a variety of cancer cell growth–related cellular processes and for development of metastasis (31–35). Selective inhibition of farensyl pyrophosphate synthase (FPPS) by bisphosphonates can lead to diminished posttranslational prenylation of small GTPase proteins (including signaling molecules such as proteins of the Ras and Rho families) which promote tumorigenesis and metastases (36). Inhibition of angiogenesis and of tumor cell adhesion as well as promotion of apoptosis are other antitumor mechanisms that have been suggested for bisphosphonates (28, 37, 38). Use of bisphosphonates was shown in preclinical studies to be associated with the suppression of Rho and Ras pathways (37). Statins, which block the mevalonate pathway, as do bisphosphonates, by acting upstream of the farensylation/geranylation step, have also been shown to be associated with reduced risk of cancer in the colon (39) and other tumor sites and with possibly improved survival of patients with cancers in multiple sites (40).

We studied a large cohort of women with breast cancer to assess the association between bisphosphonates use and survival.

Participants in this analysis come from a population-based cohort of consecutively diagnosed patients with breast cancer. These were recruited within the framework of ongoing case–control study in Northern Israel. Patients who lived in a geographically defined area of Northern Israel at time of diagnosis were eligible to participate. The Breast Cancer in Northern Israel Study (BCINIS) is a population-based case–control study of incident female breast cancer in Northern Israel with recruitment starting at January 1, 2000 (13). Participants provided written informed consent at the time of enrollment and were interviewed to obtain information about their personal and family history of cancer, reproductive history, medical history, medication use, and health habits including a food frequency questionnaire. Venous blood samples were drawn and paraffin-embedded tumor blocks of tumor tissue were sought for all cancer cases. Included in this analysis are women diagnosed with cancer in the breast who reported being postmenopausal at time of diagnosis or were older than 55 years and who were Clalit Health Services (CHS) insurees (for whom computerized full prescription data were available). CHS is the largest health care provider in Israel and covered, during the study years, approximately 70% of the older population (persons 55 years or older). Health care coverage in Israel is mandatory and is provided by 4 groups akin to not-for-profit health maintenance organizations. Thus, all study participants had a similar health insurance plan and similar access to health services, including prevention, cancer screening, and oncologic treatments. The institutional review board at the Carmel Medical Center, Haifa, Israel, approved all procedures.

The use of bisphosphonates was determined on the basis of CHS pharmacy records that were available for all CHS members study participants included in this analysis. These records included all filled prescription from the year 1998 to current and could therefore be separated into medications used before and after cancer diagnosis. Detailed prescription information enabled us to evaluate the type of bisphosphonate used, as well as the length of use and the number of prescriptions filled. Because of the very low co-payment required when filling the prescription within the Clalit health system, it is unlikely that prescription medications were purchased in private, non-CHS, pharmacies. For this study, bisphosphonate users are defined as women whose alendronate and/or risedronate use was only after the diagnosis of breast cancer and before the index date (see Follow-up Data for definition). Women who used zolendronate at any time were excluded from the study.

Vital status and other demographic parameters, such as year of birth, sex, country of birth, and ethnic group, were available for all study participants through August 31, 2014. Cause of death was available for the majority of those who died (77%) from either the Central Bureau of Statistics or from medical records of the deceased that were extracted by the research team. When the assigned cause of death conflicted between the sources, priority was given to the data extracted from the medical records. Histological diagnosis and data on stage and grade at diagnosis were available for the majority of participants (67%–87%). Data were missing mostly in women who did not undergo surgery due to age or advanced disease at diagnosis. Missing pathologic grade was usually limited to specific histologic subtypes in the breast study. Hormone receptors [estrogen receptor (ER), progesterone receptor (PR)] and Her2neu status were available for 83% to 95% of participants in the breast cancer cohort. The main outcome of this study was all-cause mortality.

A nested case–control study from within the cohort described above was used to deal with the time-varying onset of drug use and different lengths of follow-up and drug use times. For each case of death (cases), up to 30 patients with breast cancer (controls) were randomly selected from the cohort using risk set sampling (41). The controls were matched by date of entry to the cohort and year of birth (±2 years). Controls had to be alive at the time of death of the case; therefore, each control had the same follow-up time as its matched case. The index dates of the controls were the dates corresponding to the same follow-up days as their matched cases, such that follow-up days were equal. Use of bisphosphonate in controls was cutoff at the index date of the case. When fewer than 30 controls were found, all available matches were used. The same control could have been randomly selected numerous times; thus, the number of controls is much larger than the number of cases.

Statistical analyses were performed using SPSS (v21.0) and SAS version 9.3. Conditional logistic regression was used to compute the ORs of death of any reason and for death from breast cancer associated with bisphosphonate use, as the OR gives an unbiased estimate of the rate ratio in a nested case–control study (42).

The medical records of 3,731 breast cancer cases that were postmenopausal at diagnosis and were members of CHS served as the source for the study population. All eligible women were previously unexposed to bisphosphonates before their cancer diagnosis and had a median follow-up time of 70 months. Among these 3,731 cases, 481 were users of second-generation bisphosphonates, with a median of 27 filled prescriptions (1 prescription = 1-month use) during the follow-up period.

There were 799 deaths during a mean follow-up time of 70 months. Of these 799 cases, 773 were matched to 15,915 “controls” from the same pool of breast cancer patients (no controls were found for 26 cases). Of the 773 included cases, 336 (43.5%) died of breast cancer, 50 (6.5%) died of other malignancies, 207 (26.8%) of nonmalignant causes, and the cause of death of 180 (23.3%) women was unknown.

The median number of controls matched per case is 12; the median number of times the same control was used is 4; 95.5% of potential controls were chosen at least once.

In a separate analysis, 7,749 controls were matched to the 336 cases who died of breast cancer.

Table 1 presents the demographic and clinical characteristics of all cases (all-cause mortality) and their matched controls. The mean age of the menopausal women in our study was 63.3 years. Almost 7.2% of the cases and 9.4% of the controls ever used second-generation bisphosphonates. As expected, the controls had more favorable clinical characteristics (stage, grade, receptors) than the cases (who by definition died), but the difference in clinical characteristics between cases who used bisphosphonates and those who did not was much smaller and many times nonsignificant.

In this breast cancer cohort, previously unexposed women who took bisphosphonates for more than 18 months after diagnosis demonstrated significantly decreased mortality compared with women who did not take bisphosphonates or took them for <18 months (OR, 0.58; 0.38–0.88; P = 0.01; Table 2) This advantage remained significant after adjusting for tumor stage and grade(OR, 0.63; 0.41–0.96; P = 0.03). When restricting to deaths only due to breast cancer, the crude OR was 0.44 (0.20–0.95; P = 0.04); after only 12 months of use and the stage, grade-adjusted OR was 0.40 (0.18–0.90; P = 0.025; 18-month data: OR, 0.28; 0.09–0.91; P = 0.035). Exclusion of women who were diagnosed with metastatic disease at presentation did not materially change the results. Similar point estimates for the ORs of people dying were achieved for ER-positive breast cancers, ER-negative tumors, triple-negative tumors, and HER2-positive tumors comparing users and nonusers of bisphosphonates, but the study did not have enough power to reach statistical significance of these estimates.

Among postmenopausal women diagnosed with breast cancer, the use of second-generation oral bisphosphonates for at least 18 months was strongly and significantly associated with significantly less mortality. The use of bisphosphonates after diagnosis was associated with both reduced overall mortality and cancer-specific mortality by 47% to 72%.

Bisphosphonates have already been shown in case–control studies to be associated with reduced risk of developing breast and colon cancer and to lead to reduced risk of contralateral breast cancer (43). In randomized trials employing third-generation intravenously administered bisphosphonates, bisphosphonates reduced the rate of bone metastases (6, 24) and improved survival in pre- and postmenopausal women with breast cancer (22, 23). Animal models have also demonstrated a significant antitumor activity of a variety of bisphosphonates (44, 45).

An array of mechanisms for a putative anticancer activity of bisphosphonates has been suggested. Proposed mechanisms involve bone turn over processes (30), as well as suggested activity which is not bone-mediated, such as blockade of the mevalonate pathway, leading to reduced cell proliferation and reduced production of Ras and Rho proteins (46, 47).

Our current results further suggest an overall survival advantage of using oral bisphosphonates after diagnosis of breast cancer. This is of even more interest because most women detected with nonmetastatic breast cancer that later die, end up dying of causes other than the breast cancer itself.

Our data are based on a large cohort of consecutively diagnosed cancer cases with a long follow-up period of a median of 5.8 years. Extended follow-up allows for a substantial event rate even for tumors with relatively good prognosis such as early-stage breast cancer. The mean age of the women participating in the breast cancer study was 63.3 years. At these ages, a substantial proportion of the cases are expected to die of causes that are not cancer. We had good and full access to vital status data and the number of women lost to follow-up. Women with unknown cause of death were still accounted for in the overall mortality analysis and were unlikely to have died of cancer. Restricting the data to only women who are members of CHS was done to increase the validity and depth of information regarding the use of bisphosphonates (type, duration, amount) and is not expected to lead to a foreseeable selection or generalizability bias, as all women in Israel are insured with the same health services basket and are being cared for in the same oncology centers. Information bias is also unlikely, as CHS members receive their medication at CHS pharmacies and have no positive incentive to purchase them privately. All CHS pharmacies are computerized to a central database since 1998.

The role of bisphosphonates in premenopausal women or in men with cancer cannot be stated at this stage before evidence is collected specifically in these sub-groups.

In summary, oral bisphosphonates, if used in previously unexposed women for at least 18 months after cancer diagnosis, appear to improve both the odds of surviving breast cancer and the odds of overall survival.

No potentials conflicts of interest were disclosed.

Conception and design: G. Rennert, G. Fried, A. Bitterman

Development of methodology: G. Rennert, W. Saliba, K. Landsman, H.S. Rennert

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Flugelman, H. Goldberg, G. Fried, M. Steiner, A. Bitterman

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): G. Rennert, N. Gronich, W. Saliba, I. Lavi, K. Landsman, H.S. Rennert

Writing, review, and/or revision of the manuscript: G. Rennert, N. Gronich, W. Saliba, H. Goldberg, G. Fried, M. Steiner, K. Landsman, H.S. Rennert

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M. Pinchev, A. Flugelman, H.S. Rennert

Study supervision: G. Rennert

Funding for this study was provided in part by a grant from the Breast Cancer Research Foundation (BCRF), New York.

The funding agency, which is a public, not for profit, fund raising organization, had no role in the conduct of the study, in evaluation of the results or in the decision to publish.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.