Up to one-third of patients with early stage breast cancer will eventually die from the disease, and most of these (~80%) will have bone metastases. Although a majority of these bone metastases are destructive or osteolytic, a significant percentage also causes abnormal bone formation or osteosclerotic lesions. Once tumor has metastasized to bone, the disease is incurable. Because the average survival of breast cancer patients following diagnosis of bone metastases is 24–36 months, the morbidity of bone pain, fracture, hypercalcemia and nerve compression syndromes is longstanding. Therapeutics to treat and prevent these devastating complications of bone metastases are therefore in great demand.

It is well established that the skeleton is the most common site of distant metastases of breast cancer cells. Paget proposed, in 1889, that the affinity of certain cancers to metastasize to bone was due to the fact that the bone provides a 'fertile soil' or environment for the cells to germinate [1]. This seed and soil hypothesis is supported by the fact that bone is a repository for a number of growth factors and that osteoclastic bone resorption releases these growth factors. Histological sections of breast cancer metastases to bone reveal tumor cells adjacent to osteoclasts that are resorbing bone. These observations, combined with the clinical data demonstrating that bisphosphonate inhibitors of bone resorption reduce skeletal morbidity in breast cancer patients, indicate that bone destruction in breast cancer osteolysis is mediated by the osteoclast.

Our laboratory and other laboratories have provided evidence of a 'vicious cycle' involving breast cancer and bone. In this vicious cycle, metastatic breast cancer cells in bone produce factors (such as parathyroid hormone-related protein) that stimulate osteoclastic bone resorption. This production results in the release of growth factors, such as transforming growth factor-β, from the bone matrix [2, 3]. Growth factors, in turn, stimulate tumor growth and production of more parathyroid hormone-related protein, resulting in a 'vicious cycle' that further fuels the bone destruction and tumor growth [2, 3].

These local tumor–bone cell interactions resulting in osteolysis are the final steps of the journey that a tumor cell navigates from the primary site to the skeleton. The tenacious tumor cell must undergo the multistep process of the metastatic cascade before it comfortably settles in bone. Specifically, the tumor cell must detach from the primary site, enter tumor vasculature to reach the circulation, survive host immune response and physical forces in the circulation, arrest in a distant capillary bed, escape the capillary bed, and proliferate in the metastatic site.

The events involved in entering the tumor vasculature are similar to those involved with exiting the vasculature in the bone marrow cavity. These events include attachment of tumor cells to the basement membrane, tumor cell secretion of proteolytic enzymes to disrupt the basement membrane, and migration of tumor cells through the basement membrane. Attachment of tumor cells to basement membranes and to other cells is mediated through cell adhesion molecules such as laminin and E-cadherin. Tumor cell secretion of substances such as metalloproteinases facilitates disruption of the basement membranes and enhances invasion. Inherent tumor cell motility or motility in response to chemotactic stimuli are also important factors for tumor cell invasion to the secondary site.

A number of other tumor types frequently metastasize to bone, including prostate, lung, thyroid, and renal cell carcinomata. Multiple myeloma is a primary hematologic malignancy that causes osteoclast-mediated bone destruction but is not a metastatic tumor per se. However, many of the tumor cell–bone cell interactions that cause the bone destruction are similar to those implicated in breast cancer osteolysis. Although this review focuses on breast cancer metastases to bone, the studies reviewed here offer general mechanisms that can be applied to other tumor types.

Bisphosphonates are synthetic analogs of inorganic pyrophosphate. They are taken up preferentially by the skeleton and are strongly bound to hydroxyapatite on the surface of bone. Bisphosphonates are potent inhibitors of osteoclastic bone resorption. The effects of these drugs are primarily on the bone-resorbing osteoclasts but may also target osteoblasts, macrophages and tumor cells [4–8]. The mechanisms by which bisphosphonates inhibit osteoclast activity, and the relative potencies with which they do so, are dependent on the molecular structure of each compound.

The major mechanism of the nitrogen-containing bisphosphonates to decrease osteoclast number, osteoclast activity and bone resorption is by induction of osteoclast apoptosis [9]. As described by Fleisch in this issue [10], bisphosphonates inhibit farnesyl diphosphate synthase in the mevalonate pathway and thereby prevent protein prenylation of small GTPase signaling proteins required for osteoclast function [11, 12]. The degree to which nitrogen-containing bisphosphonates inhibit bone resorption correlates with the capacity to cause apoptosis in cells of the osteoclast lineage, as well as with the capacity to inhibit farnesyl diphosphate synthase and protein prenylation in the osteoclast [11]. The non-nitrogen-containing bisphosphonates, such as clodronate, do not inhibit protein prenylation and have a different mode of action that may involve the formation of cytotoxic metabolites in osteoclasts or inhibition of protein tyrosine phosphatases [12].

Bisphosphonates also affect cells other than osteoclasts in the bone microenvironment. Derenne et al. showed that the bisphosphonates pamidronate and zoledronate inhibited interleukin-6-induced production of matrix metalloproteinase-1 by bone marrow stromal cells [13]. In vitro studies have also shown that nitrogen-containing bisphosphonates inhibit the adhesion of breast cancer cells to bone and bone matrix [14].

Even more intriguing are those results indicating that bisphosphonates may have direct effects on the tumor cells themselves, as reviewed in detail by Senaratne and Colston in this issue [15]. Bisphosphonates have been shown to decrease the proliferation and viability of human tumor cells lines, as well as to increase the apoptotic index of the human tumor cell lines [13, 16]. Numerous investigators have demonstrated that bisphosphonates significantly reduce tumor invasion and adhesion to bone [13, 14, 16–19].

Studies of bisphosphonates in preclinical models of breast cancer metastases to bone illustrate the importance of the bone microenvironment and osteoclastic bone resorption in the development of skeletal metastases of both osteolytic and osteoblastic nature. We have learned from these studies that the primary action of bisphosphonates is on the bone resorbing osteoclasts, and that bisphosphonates may exert secondary effects on the tumor cells in bone. Zoledronate, the newest and most potent bisphosphonate, showed promise in the preservation of bone in the 4T1 model of breast cancer metastasis to bone [23]. It caused a decrease in the osteoclast number and a decrease in the tumor burden in bone [23].

There are multiple explanations for this decrease in tumor burden in bone. First, and most probable, the decrease in osteoclastic bone resorption makes the skeleton a less hospitable environment for the tumor cells, by reducing the release of bone-derived growth factors that may stimulate tumor growth or production of osteolytic factors. Second, the bisphosphonates may have a direct effect on the tumor cells to induce apoptosis. While there are data in vitro and in animal models to support this, the concentrations of bisphosphonates at which this occurs in vitro are quite high. It is unclear whether such concentrations occur in vivo. Such data have yet to be demonstrated in humans with bone metastases. Finally, it is also possible that the decrease in tumor burden in bone is due to an effect of bisphosphonates on angiogenesis [35]. This potential anti-angiogenic effect on tumors remains to be investigated.

The observation in some of the animal models reported here that clearly needs to be clarified in human studies is the issue of extraskeletal metastases. That is, soft tissue metastases were increased following treatment with bisphosphonates, an observation indicating that if the tumor cells find the bone microenvironment inhospitable then they may seed to other tissues. However, clinical data with regard to this are inconsistent. Diel et al. found a decrease in visceral metastases in patients with breast cancer following treatment with clodronate [36], while Saarto et al. found an increase in visceral metastases [37].

Many questions remain regarding the use of bisphosphonates in the treatment of metastatic breast disease to the skeleton. More work is needed to determine whether bisphosphonates truly have antitumor activity in humans, or whether the reduction in tumor burden in bone is due to the reduction of bone-derived growth factors released into the local microenvironment as a consequence of inhibiting osteoclastic bone resorption. Second, bisphosphonate use in osteoblastic metastases needs to be further explored to prove definitively that treatment inhibiting bone resorption does indeed reduce osteoblastic disease. In addition, it still remains to be determined whether bisphosphonates increase survival in clinically relevant situations. Patients with metastatic breast cancer are treated with anticancer agents in addition to bisphosphonates, and studies in animal models that mimic this situation would provide more realistic evidence with regard to survival. Finally, the use of bisphosphonates in the treatment of bone metastases of other primary tumors is an area that needs to be explored in both animal models and humans. Many of these questions are currently under investigation, and the answers should provide a strong rationale for the use of bisphosphonates in cancer-induced morbidity of the skeleton.