Targeting the endothelin receptor in prostate cancer bone metastasis: back to the mouse?

Prostate cancer is the most common malignancy in American men. Although mortality associated with prostate cancer has been declining over the past decades, complications related to metastasis remain a major challenge. Prostate cancer cells have a tendency to metastasize to bone, disrupting normal bone homeostasis and negatively impacting patient morbidity and mortality. For these reasons, there are ongoing efforts aimed at developing strategies for specific targeting of prostate cancer bone metastases. The vasoactive peptide endothelin-1 (ET-1) has emerged as an important player and a potential therapeutic target with a role in prostate cancer bone metastasis ¹.

ET-1, together with ET-2 and ET-3 are 21-aa peptides encoded by distinct genes that form the endothelin family. Endothelins exert their effects through G-protein-coupled endothelin receptors A and B (ETA, ETB). ET-1 acts primarily through ETA to activate a signaling pathway that can be modulated by activation of receptor tyrosine kinases such as the epidermal growth factor receptor and the erbB receptor ^{2, 3}. The ET-1 axis is involved in several aspects of cancer progression including proliferation, apoptosis, angiogenesis, and bone metastases ^{1, 4–6}. Several studies indicate that the ET axis is involved in the pathogenesis of prostate cancer. ET-1 levels are elevated in prostate cancers ^{7, 8}, as are levels of the ETA receptor in proportion to tumor grade and stage ^{1, 7}. Due to the intrinsic interactions between metastatic tumor cells and the bone environment (e.g. osteoblasts and osteoclasts) and the prominent role of ET-1 in bone metastases, substantial efforts have been aimed at evaluating antagonists targeting the ET-1 axis. In particular, a selective ETA antagonist, atrasentan (ABT-627), has been the focus of preclinical and clinical trials. In mouse models, atrasentan has been shown to prevent osteoblastic bone metastases ⁹ and to potentiate the effects of the chemotherapeutic agent docetaxel ¹⁰.

Despite these encouraging preclinical studies however, clinical trials with atrasentan in prostate cancer have been inconclusive. In the initial phase I trials conducted in patients with castration-resistant prostate cancer (CRPC), both cancer-related pain and PSA levels were reduced by atrasentan ¹¹. In phase II trials conducted in patients with CRPC and evidence of metastatic disease, disease burden was attenuated and median time to PSA progression was also delayed by atrasentan treatment ¹². Furthermore, atrasentan suppressed biochemical bone turnover and showed some evidence of slowing metastatic cancer progression in bone ¹³. Despite these initial successes however, phase III clinical trials in patients with both metastatic and non-metastatic prostate cancer have been somewhat disappointing. Atrasentan did not significantly delay time to disease progression or overall survival, although it had positive effects on time to PSA progression and the level of serum markers of bone turnover ^{14, 15}. Nevertheless, the disappointment associated with the outcome of these phase III trials has been tampered by several considerations. These include the possibility that radiographic assessment of progression could have been a confounding factor in the study ¹⁴ and a possible inappropriate use of PSA levels to drive the decision to prematurely discontinue the drug in some cases ¹⁶. There is also some hope from pre-clinical studies that combinations of atrasentan with other agents such as docetaxel will be more successful ^{10, 17, 18}.

Regardless, there is the need to fully understand the mechanism of action and tumor stage/site affected by atrasentan to aid in the correct design of future clinical trials. This requires the use of improved preclinical in vivo models coupled with careful analysis. In this regard, the study by Drake et al in this issue of Cancer Biology and Therapy ¹⁹ is a welcome step in the right direction. Considering that most current models of ET-1 action in prostate cancer suggest it will have a specific effect on bone metastasis, these investigators performed a serial analysis of the fate of bone versus soft tissue metastases in a prostate cancer model after treatment with atrasentan. They used a previously developed mouse model of prostate cancer metastases following intra-cardiac injection of luciferase-expressing 22Rv1 human prostate cancer cells ²⁰. The fact that 22Rv1 cells do not express appreciable levels of the ETA receptor or respond to ET-1 in vitro ²¹ allowed the investigators to separate any effects of atrasentan on the tumor cells from effects on the tumor microenvironment. Using bioluminescence imaging (BLI) to monitor metastatic tumor colonization and progression, Drake et al noted an interesting phenomenon: while the overall tumor growth and survival of mice were not significantly affected by atrasentan, the drug clearly had an effect on the subset of mice with bone metastases. Atrasentan specifically inhibited growth of bone metastases but not metastases in soft tissues such as liver and adrenal gland even in the same animal.

Since 22Rv1 cells do not appear to be sensitive to atrasentan, by what mechanism might the drug be working? The ability to follow tumor growth kinetics in vivo by imaging proved to be an advantage here. Drake et al found that atrasentan did not inhibit the initial colonization and growth of the tumor cells in bone for up to about 4 weeks. However, after the bone metastases have reached a certain size, there is a rapid collapse of bioluminescence signal, indicative of tumor regression. The fact that 22Rv1 cells express very little ETA receptor and are insensitive to the cell growth inhibitory effects of atrasentan in vitro makes it unlikely that tumor regression was achieved by a direct effect of atrasentan on 22Rv1 tumor cells unless these cells gained new characteristics in bone environment. Bone metastasis is a dynamic process where tumor cells and bone environment actively interact and the possibility that the 22Rv1 cells might have acquired the ability to express ETA receptor was not formally ruled out.

A more likely -albeit speculative- hypothesis is that atrasentan caused tumor regression by disrupting tumor-bone interactions, especially tumor-osteoblast interaction, ultimately affecting angiogenesis (Figure 1). The endothelin axis in osteoblasts could influence the production of pro-angiogenic molecules such as VEGF. The latter can then promote angiogenesis which feeds tumor growth. Interruption of this paracrine loop by atrasentan acting on osteoblasts is expected to ultimately block angiogenesis and lead to tumor regression. Although ET-1 can directly modulate angiogenesis via ETB receptor present on endothelial cells ⁵, a direct effect of atrasentan on endothelial cells is unlikely given the reduced selectivity of atrasentan to ETB. Since vascular smooth-muscle cells and pericytes express the ETA receptor and are stimulated to grow by ET-1, these cells also might be targets of atrasentan ^22–25.

Figure 1.

Cartoon depicting a hypothetical model of the effect of atrasentan treatment on 22Rv1 prostate tumor bone metastasis. Atrasentan is proposed to interfere with tumor-bone interactions, especially tumor-osteoblast interaction. This will affect osteoblast function, e.g. the production of pro-angiogenic molecules such as VEGF, blocking angiogenesis which is required for sustained tumor cell growth and this can ultimately lead to tumor regression. Other possible targets include osteoclasts, vascular smooth-muscle cells and pericytes, all of which may contribute to tumor cell growth.

Although the study by Drake et al has made some important observations, it has also raised several unanswered questions and comes with some caveats. First, while the lack of expression of ETA receptor by 22Rv1 cells is an advantage experimentally, the ET-1 axis is active in many tumor cells in prostate cancers and prostate cancer bone metastases. Thus the response of the tumors in vivo to ET-1 blockade might differ from that observed in the Drake et al study. Second, the study did not include a detailed histological analysis of the metastatic lesions to shed light on the cellular and molecular processes associated with tumor regression in treated animals. Third, the study involved a single cell line, 22Rv1, which produces mixed osteolytic/osteoblastic bone lesions. These may differ in their response to atrasentan from metastases with a predominantly osteoblastic phenotype. These and other questions should form the subject of interesting future research.

Overall, this study makes a number of important points. ETA receptor blockade could be an effective treatment for prostate cancer bone metastases, but not for non-bone metastases. Even for bone metastases, initial stages of tumor colonization and growth may not be affected by atrasentan. This emphasizes the need to choose appropriate biomarkers and endpoints in clinical studies evaluating targeted therapies. Another important lesson worth noting is how, despite some of their limitations, the appropriate use of mouse models could help inform the proper design of clinical trials and perhaps use of therapeutic agents. Therefore, even for agents that have undergone some clinical testing, sometimes it may be worthwhile to consider going back to the mouse.

Abbreviations

ET-1

Endothelin-1

ETA

Endothelin receptor A

ETB

Endothelin receptor B

CRPC

Castration-resistant prostate cancer

PSA

prostate-specific antigen

BLI

bioluminescence imaging

VEGF

Vascular endothelial growth factor