Endothelin-1 and osteoblastic metastasis

When cancer metastasizes to distant sites, it essentially forms an army of occupation. Just as the presence of an army of occupation alters the behavior of the preexisting resident population, the resident host cells react in a variety of ways to the presence of the metastatic cancer cells. In the majority of cases, they increase their resorptive or destructive capacity, and the result is an osteolytic bone lesion. However, in some notable cases, the response is predominantly formative or osteoblastic. This is particularly true for metastatic prostate cancer, although it is also frequent in a variety of hematologic malignancies as well as in patients with breast cancer. The mechanisms responsible, until recently, have been largely unknown and undefined.

It is well known and widely lamented that animal models that consistently produce prostate cancer-induced osteoblastic metastases are virtually nonexistent. Recently, however, there have been successful efforts to use breast cancer models to reliably produce these lesions (1, 2). The study by Yin et al. (3) in this issue of PNAS also uses a preclinical model of osteoblastic metastasis caused by human breast cancer cells to show that endothelin-1 is the responsible mediator for these lesions in this case, and that specific antagonists to the endothelin-1 receptor inhibit the osteoblastic response to the presence of tumor in the bone marrow cavity. The model comprises nude mice inoculated with human breast cancer cells in the left cardiac ventricle, after which the cancer cells invade bone marrow and other distant sites to form metastases. The particular human tumors studied by Yin et al. formed osteoblastic metastases, although other human breast tumors form osteolytic or mixed osteolytic–osteoblastic metastases under these circumstances (1, 2, 4).

These observations represent a major step forward in our understanding of the potential mechanisms involved in the pathophysiology of osteoblastic metastasis. In the past, many mechanisms for this phenomenon have been proposed, including production by tumor cells of growth regulatory factors such as the fibroblast growth factors and bone morphogenetic proteins, and fragments of plasminogen activator and of PTH-related peptide (PTH-rP) (5–8). However, none of these factors has been proven to be causative in any existing model. What these authors have shown is that the tumor they studied expressed endothelin-1, that conditioned media from the tumor cells stimulated bone formation in vitro through an endothelin-1-mediated mechanism, and that inhibition of endothelin-1 by the use of a specific receptor antagonist reduced the bone lesions in vivo. This is very convincing evidence, similar to that which showed that PTH-rP was responsible for humoral hypercalcemia of malignancy (9), and that local production of PTH-rP was responsible for osteolytic metastasis caused by human breast cancer cells (2).

These observations are a major step forward in our understanding of osteoblastic metastasis.

There are several other insights with respect to osteoblastic metastases that come from this study. One interesting aspect is that administration of this receptor antagonist to nude mice inoculated with the breast cancer cells led not only to reduction of the bone lesions, but also to a reduction of the tumor burden that was present in the bone marrow cavity. There was no increase in the metastases in visceral organs, always a theoretical concern when agents are used that target bone metastases. This finding is reminiscent of what has been described when similar preclinical models of osteolytic metastasis are treated with bisphosphonates (1), neutralizing antibodies to PTH-rP, the causative factor in that situation (2), or drugs that inhibit PTH-rP transcription (10). In these cases, the treatment intervention reduces the bone lesion either directly, by inhibiting osteoclast function (bisphosphonates), or indirectly, by inhibiting the factor responsible for osteoclast activity (neutralizing antibodies to PTH-rP or PTH-rP transcriptional inhibitors). As a consequence of the decreased osteoclastic bone resorption, there is reduced osteolysis, as well as a decrease in tumor burden. This same phenomenon has occurred in the experiments reported by Yin et al. The reduction in the bone lesions suggests, as is the case for osteolytic metastasis, that there is a “vicious cycle” between the bone microenvironment and the metastatic cancer cells, such that the cancer behavior in the bone microenvironment is driven by the presence of the bone lesion. In addition, treatment of the bone lesions leads to improvement not only in the bone-related effects, but also in the tumor burden present in the marrow cavity. This “vicious cycle” concept emphasizes the importance of the bidirectional interactions that occur between tumor cells and bone cells at the metastatic site and that drive the metastatic process. The analogy of the army of occupation and its dependence on a cooperative host population holds here as well.

Another interesting aspect of this study is the potential for endothelin-1 to act as a physiologic regulator of bone formation. The pathophysiologic effects of endothelin-1 on bone formation reported here may be limited to pathologic situations, such as cancer metastasis, but they may also represent a cancer-related aberration of a normal physiologic mechanism, which poses an intriguing question. If this does represent a physiologic mechanism gone awry when driven by malignancy, then is it possible that this receptor antagonist or similar pharmacologic agents designed to block the endothelin-1 signal transduction pathway may have pharmacologic applications in other bone diseases not associated with malignancy? If it turns out that endothelin-1 is a regulatory factor involved in normal bone remodeling, then the possibility that this or related compounds have other applications becomes even more likely. Future studies will be important in exploring this possibility.

Endothelin-1 is not the only stimulator of bone formation that has recently been linked to osteoblastic metastasis. In a similar model, but using different human breast cancer cells, Yi et al. (4) have shown that platelet-derived growth factor (PDGF-BB) was the responsible mediator of osteoblastic metastasis. This was demonstrated by showing increased expression of PDGF-BB by metastatic human breast cancer cells and inhibition of its effects both in vitro and in vivo with antisense oligonucleotides to PDGF-BB. It remains to be shown how relatively frequent PDGF-BB and endothelin-1 are as mediators of osteoblastic metastases, or even what relationship these two factors have, if any. The PDGF-BB tumor cells caused similar osteoblastic metastases to those described by Yin et al., although in the former case there appeared to be an important associated resorptive element. The significance and cause of the resorptive component and the relationship of this resorptive component to PDGF-BB remain unknown.

The role of bone resorption in the process of osteoblastic metastasis requires further investigation. It is now apparent that, in most cases of osteoblastic metastasis, increased bone resorption plays an important role in the process, although this was not apparent in the model described by Yin et al. In the majority of patients, markers of bone resorption are increased (11–13), and drugs that specifically inhibit osteoclastic activity, such as the bisphosphonates, relieve pain and reduce skeletal-related events (14). Whether this increase in bone resorption is an essential precedent for the osteoblastic response or a consequence of the osteoblastic lesions remains unknown. However, preliminary evidence suggests that bisphosphonate treatment reduces not only bone pain and skeletal-related events but also tumor burden. If this is so, it suggests that the bone resorption component does indeed precede the osteoblastic response.

Our understanding of the process of osteoblastic metastasis has been severely impaired by the lack of good preclinical models. The model described by Yin et al. is an attractive preclinical model because it utilizes human tumor cells associated frequently with osteoblastic metastases, and delivers these tumor cells to the bone metastatic site via the blood stream, as is the case in naturally occurring bone metastases. Other investigators have used direct inoculation of tumor cells into the bone site (15), but this always involves local injury to bone, which can hamper interpretation of results. As noted previously, what is badly needed in this field is a model of prostate cancer metastasis to bone causing new bone formation at the metastatic site, because prostate cancer is the human malignancy most often linked to osteoblastic metastases. None of the human prostate cancer cell lines currently available reliably causes osteoblastic lesions in animal models. Rather, they typically cause osteolytic lesions because of the overproduction of PTH-rP. It is unknown why it has remained so difficult to develop a useful prostate cancer model of osteoblastic metastasis. This notwithstanding, it is likely that the mechanisms shown here by Yin et al. will be similar in many cases of spontaneous and naturally occurring clinical human prostate cancer metastasis, because Nelson et al. previously showed that there are increased circulating concentrations of endothelin-1 in patients with metastatic prostate cancer cells (16). These results beg future study to identify and evaluate new models of prostate cancer metastasis.

Endothelin-1 may be a physiologic regulator of bone remodeling.

The observations made by Yin et al. are very important. They not only increase our understanding of the mechanisms involved in osteoblastic metastasis, but also point to the potential of endothelin-1 as a physiologic regulator of bone remodeling and show that its effects can be neutralized by specific receptor antagonists. They also show the potential for excellent preclinical models to influence our thinking about pathophysiologic mechanisms, as well as how such information can be used to inhibit the pathophysiologic process. It is also hoped that studies using models such as this can beneficially influence the design of clinical trials to test various agents either solely or in combination in the treatment of cancers metastatic to bone. This is not simply an issue of academic interest, because bone metastasis is a very important and frequent clinical problem. Indeed, each person has approximately a one in five or one in six chance of developing a bone metastasis during the course of their lives. These current data show that this particular variant of cancer metastasis might be successfully treated by what appears to be a relatively safe agent with the goal of reducing the bone effects and, maybe even more importantly, reducing the tumor burden in the bone marrow cavity. Studies such as those of Yin et al. reported here may have far-reaching consequences for patients with this common and devastating manifestation of common metastatic malignancies.