Targeting the premetastatic niche

In collaboration with David Lyden (Cornell University Weill Medical College), we have discovered that nonmalignant bone marrow cells establish "cellular bookmarks" in target organs that guide cancer cells to their predetermined destination. This finding could have a major impact on how oncologists assess the likeliness of metastasis to specific organs. This discovery may also help identify subsets of high-risk cancer patients who are prone to distant metastases. Those patients would likely benefit from a more aggressive adjuvant therapy. Understanding how cellular bookmarking works at the molecular level could lead to new information that may help thwart metastasis, a major cause of death among cancer patients.

We have established that a specific subset of bone marrow–derived cells (BMDCs)-which are composed of hematopoietic progenitor cells capable of dividing and forming colonies-are recruited by tumors to aid in the growth of new blood vessels. The generation of new blood vessels occurs through a process called angiogenesis. In previous studies, we had shown that corecruitment of hematopoietic BMDCs expressing the angiogenic factor receptor VEGF-R1, along with the vascular cells, accelerated the assembly of newly formed blood vessels and tumor growth.

We have demonstrated that a nonmalignant cluster of VEGF-R1 hematopoietic BMDCs is recruited to a premetastatic niche, thereby establishing a docking site prior to the arrival of the circulating tumor cells.

In experiments with mice that had been implanted with highly metastatic lung cancers or melanoma cells, we discovered that BMDCs do, indeed, arrive at the premetastatic sites before the arrival of cancer cells. We also found that such clusters appear prior to the development of metastases in mice genetically predisposed to developing tumors-a system that closely mimics how cancers develop.

Interference with the mobilization of VEGF-R1 cells from the bone marrow and incorporation into the premetastatic niche significantly decreased subsequent tumor metastasis. Moreover, depleting VEGF-R1 cells or inhibiting the function of VEGF-R1 itself also retarded the spread of tumors to their predestined metastatic sites. Remarkably, tumor types determined the pattern of organ localization of BMDCs. By releasing soluble factors, tumor cells were directing BMDCs to spread to the sites where they were supposed to go. For example, melanoma cells that have the capacity to metastasize to virtually every organ released factors that directed incorporation of VEGF-R1 BMDCs to all of the organs that are known to be the common sites for melanoma metastasis.

We also identified a number of regulatory molecules, including the adhesion molecule VLA4 and the protease MMP-9, which are necessary for BMDCs to establish the premetastatic niches in target organs and for tumor cells to find and attach to those niches. VLA4 enables the attachment of BMDCs to components of tumor stroma, such as fibronectin. Soluble factors released by the tumor cells selectively stimulated the production and deposition of a matrix molecule, fibronectin, which provided a docking site for the attachment of BMDCs before the arrival of the tumor cells.

In another clinically relevant observation, we found numerous VEGF-R1 clusters in various tissues obtained from patients with breast, lung, and esophageal cancers. Conventional diagnostic techniques, such as light microscopy, may fail to detect very small micrometastatic tumors in the lymph nodes in the immediate vicinity of the primary tumor. But the presence of VEGF-R1 clusters might indicate undetected micrometastases, or impending metastasis, which would suggest that these particular patients may be at higher risk for tumor metastasis and thus should be treated more aggressively with adjuvant chemotherapy.

These findings also suggest that tumor metastatic potential may not only depend on the oncogenicity of the cancer cells but also on the existence of developmentally "premetastatic niches" or "hot spots" in the body that are receptive to metastatic cells.

It is conceivable that the number and capacity of these hot spots to permit the attachment of tumor cells may be determined by the genetic makeup of any given patient. For example, the propensity and magnitude of incorporation of VEGF-R1 into various organs may differ from one patient to another and might explain why subsets of patients with early-stage colon cancer are more prone to liver metastasis, while others with an identical stage of cancer and oncogenic repertoire are cured of their disease with timely surgery and adjuvant chemotherapy.

When oncologists diagnose a tumor at an early stage they face a dilemma about what to do after the surgeon has removed the tumor, particularly when pathological examination does not show any evidence of microscopic metastasis. For example, on average, only 30 percent of patients with fully resected primary tumors may relapse, while others are most likely cured of their disease. It is unnecessary to expose patients at low risk for relapse to high doses of toxic chemotherapy, which is usually associated with significant morbidity. There is a possibility, however, that the presence of VEGF-R1 hematopoietic BMDC clusters in the resected "tumor-free" tissues portends a poor prognosis; these patients may benefit from treatment with aggressive chemotherapy.