Pathogenesis and vascular integrity of breast cancer brain metastasis

Abstract

Dogma dictates that brain metastasis originate from the proliferation of extravasated tumor cells and that the blood–brain barrier (BBB) prevents the delivery of chemotherapeutic drugs to the tumors. The purpose of this study was to clarify the relationship between tumor localization and progression and the involvement of BBB function in a murine model of breast cancer brain metastasis. Green fluorescent protein expressing MDA-MB435 breast cancer cells were injected into the left ventricle of nude mice. At various time points, the entire vasculature was labeled with rhodamine-conjugated albumin. The tumors and vasculature were then imaged by laser-scanning confocal and stereo fluorescence microscopy. About 75% of the cells that reached the brain extravasated and grew perivascularly. Twenty five percent of the cells, however, proliferated within the vasculature and ultimately led to thrombosis-like infarction of the brain parenchyma. The tumorigenic “embolus” served as a sustained release source of tumor cells to downstream sites. Continuing intravascular tumor expansion led to disruption of the BBB and to overflow of cells that progressed along the vessels perivascularly to distant sites that regained protection of the BBB. Breast cancer brain metastases involve both extravascular and intravascular growth of tumor cells. These distinct pathways contribute to different pathological phenotypes that generate a heterogeneous BBB that facilitates or inhibits the delivery of chemotherapeutic drugs to the tumor. © 2006 Wiley-Liss, Inc.

Brain metastases are the most frequently occurring intracranial tumors, outnumbering primary brain tumors by a factor of 10, and represent the major cause of systemic cancer morbidity and mortality.1, 2 Their incidence is increasing because of improved cancer therapy for systemic disease.1, 3 Most patients die within a few months of diagnosis since systemic therapies are ineffective within the brain.4 Our knowledge about how brain metastases develop has, for the most part, been extrapolated from observations of tumor growth in other tissues. This has led to the concept that brain metastasis develop after extravasation of metastatic tumor cells into the surrounding tissue5, 6 where they are protected by the blood–brain barrier (BBB) that prevents the delivery of therapeutic agents to the proliferating tumor.7, 8 Because the brain lacks lymphatic drainage, the most common mechanism of brain metastasis is by hematogenous spread through the arterial circulation. By injecting tumor cells into the arterial system to mimic hematogenous spread, we developed a unique breast cancer brain metastasis model that allows us to visualize the development of brain metastasis and monitor BBB function at high resolution. Analysis of breast cancer brain metastases revealed a complex pathogenesis and heterogeneity of BBB integrity. These findings provide a framework for better understanding the clinical presentation of breast cancer and BBB function in brain tumor metastasis.

Materials and methods

Mice

Female athymic nude mice were obtained from the Animal Production Area of the National Cancer Institute-Frederick Cancer Research Facility (Frederick, MD). The animals were maintained in facilities approved by the American Association for Accreditation of Laboratory Animal Care and in accordance with current regulations and standards of the United States department of Agriculture, Department of Health and Human Services, and National Institutes of Health. All the mice were used when they were 8- to 12-weeks-old.

Green fluorescent protein-expressing MDA-MB-435

MDA-MB-435 is a human breast carcinoma cell line originally isolated from pleural effusion of a patient with advanced disease. The cell line was transfected with a plasmid, pEGFP-N1, which constitutively expresses an enhanced version of green fluorescent protein (GFP) (Clontech Laboratories, Palo Alto, CA). FuGENE 6 Transfection Reagent (Roche Molecular Biochemicals, Indianapolis, IN) was used to aid the transfections. The protocol recommended by the manufacturer was followed. G418 selection at 500 μg/ml was initiated 2 days after transfection. Two to 3 weeks later, emerging colonies were examined for GFP expression by fluorescence microscopy. Those colonies with robust GFP expression were selected, pooled and expanded.

Intracardiac injection of tumor cells

Mice were anesthetized by i.p. injection of nembutal (45 mg/kg of diabutal (50 mg/ml)/saline/ethanol/propylene glycol 10/63/7/18). The anterior chest wall was scrubbed with 70% alcohol. A 30 gauge needle on a tuberculin syringe was inserted into the second intercostal space 3 mm to the left of the sternum and aimed centrally. The spontaneous and continuous entrance of pulsating blood into the transparent needle hub indicated proper positioning of the needle into the left ventricle of the heart. Tumor cells (10⁶ cells in 0.1 ml saline) were injected over a 20–40 sec period.

Selection of MDA-MB-435GFP cells that preferentially metastasized to brain

MDA-MB-435GFP cells were injected into the left ventricle of nude mice. Brain metastases were harvested when mice became moribund; tumor cells were cultured, amplified and then injected back into the left cardiac ventricle of nude mice to generate brain metastases again. After 4 selections >80% of the mice died of brain metastasis. There were no visible metastases to other organs (such as bones, lungs) within 2 months after injection.

Labeling of vasculature in vivo

Systemic labeling of vascular network was done by i.v. injection of 1 mg of rhodamine conjugated albumin (Molecular Probes, Eugene, OR).

Examination of brain metastases and function alteration of BBB

Animals were injected with rhodamine–albumin, 0.5–1 hr before the mice were killed. The brains were removed and examined by fluorescence microscopy using a Zeiss LSM 510 laser scanning confocal microscope (LSM510, Carl Zeiss, Thornwood, New York) with argon and helium–neon lasers or a stereoscope (Leica Model LZ12), equipped with narrow bandpass excitation and emission filters mounted in a selectable filter wheel. Real-time images were directly captured with an Evolution MP camera (Media Cybernetics, Silver Spring, MD). Brain metastases were also harvested for H&E staining. All the results presented here were duplicated with murine GFP-transfected 4T1 cells indicating that the reported results were not specific to MDA-MD-435GFP.

Results

Co-existence of multiple mechanisms of breast cancer metastasis—extravasation and intravascular growth

To investigate the pathogenesis of brain metastasis and the integrity of the BBB, we injected GFP-expressing MDA-MB435 human breast cancer cells (MDA-MB-435GFP) into the left ventricle of nude mice. Sequential examination of brain lesions by stereo and/or confocal microscopy was carried out at the indicated time points 0.5–1 hr after the intravenous injection of 1 mg of rhodamine–albumin to visualize the brain vasculature.

Fluorescence microscopy of intact brains recovered from 40 mice injected with tumor cells into the left ventricle 7–14 days earlier revealed multiple foci of GFP-expressing tumor cells. While many brain metastases were located in the deep parenchyma, exposing these metastatic nodules for observation destroyed the BBB and affected the morphology of brain metastatic nodules. We, therefore, elected to investigate the superficial metastases (close to meninges). Among 28 observable MDA-MB-435GFP nodules consisting of single cells or cell clusters, 22 of 28 were extravascular (Fig. 1a), indicating that extravasation was the dominant mechanism of brain metastases formation. Interestingly, the tumor cells did not grow randomly, but aligned themselves longitudinally along the blood vessels. The longitudinal alignment could be continuous or discontinuous, suggesting that the tumor cells migrated along the vasculature (Fig. 1b). When cells divided and migrated along an existing vessel, they could spread long distances within the entire hemisphere without forming significant tumor nodules (Figs. 1c and 1d). Analysis of thin sections revealed that tumor growth could either partially or completely surround the blood vessels (Fig. 1e).

Figure 1. Extravascular growth of breast cancer brain metastases. MDA-MB-435GFP breast cancer cells (green fluorescence) were injected into the left cardiac ventricle of female nude mice. At the indicated time points, the vasculature was labeled by injecting mice with rhodamine–albumin 1 hr before harvesting intact brains (red fluorescence). Perivascular growth of metastatic cells on day 7 (a), 21 (b) and 56 (c),(d),(e). Tumors were visualized by confocal microscopy (a), (b) or stereo inverted fluorescence microscopy (c), (d). (e) shows a hemotoxylin–eosin stained thin section. (c) and (d) are the same field imaged with fluorescein and rhodamine interference filters, respectively. The arrow in (d) marks an area of the vasculature that is permeable to rhodamine–albumin.

Six of 28 metastatic nodules were intravascular at weeks 1–2. Examination of the intravascular nodules revealed that they were confined to arterioles or capillaries and extended as strings that acquired the shape of the blood vessel (Fig. 2b). Since single tumor cells populated the vasculature at day 3 and earlier (Fig. 2a and many cells at day 21 seen in Fig. 1b), this observation suggests that cell division occurred within the brain vasculature. Two additional pieces of evidence support the intravascular origin of brain metastasis. First, complete occlusion of blood vessels by tumor cells was demonstrated in areas of intravascular tumor growth (Figs. 2c and 2d) by the necrosis of brain tissue and the subsequent atrophy in the surrounding tissue (Figs. 2e and 2f). The necrotic areas usually occurred deep within the brain parenchyma, the extent of which appeared to correlate with the size of the occluded blood vessel. Second, additional tumor spread occurred within intact blood vessels through the downstream release of tumor emboli (Figs. 2g and 2h). Thus, intravascular growth of colonies became a constant source of tumor emboli that spread to distant regions, causing the metastases of metastasis. This could easily lead to extensive tumor dissemination through the vasculature and damage the host tissues located in the downstream area by blocking blood flow. Indeed, in the late stages, continuous intravascular proliferation of tumor cells led to disruption of the vessel walls (Fig. 2i) with further dissemination of tumor into the parenchyma. This led to the escape of metastatic cells that migrated and grew toward adjacent blood vessels and then proliferated along the vasculature perivascularly to distant sites.

Figure 2. Intravascular growth of breast cancer brain metastasis. MDA-MB-435GFP cells were injected into the left cardiac ventricle of female nude mice. Intact brains were recovered from mice injected with rhodamine–albumin i.v. 1 hr earlier. (a) shows a single GFP-labeled tumor cell that appeared to migrate from the large vessel (asterisk) to the small capillary where it lodged and occupied the entire lumen of the blood vessel. The arrow points to the capillary bifurcation that continues down to the right and left out of the focal plane. (b) shows MDA-MB-435GFP cells confined to a small vessel without rhodamine–albumin leakage on day 7. (c) and (d) (4-fold magnification of rectangular part of c) show intravascular growth of tumor cells on day 28. (e) H/E staining of a thin-section (from the area of the black line in d) shows necrotic parenchyma and massive inflammatory cell infiltration (white arrows). (f) H/E staining of a thin-section (from the area of the black line in c) shows tumor cells confined within the vessel (white arrow). (g) and (h) show the intravascular spread of tumor cells (in the direction of the arrow) under fluorescence and bright-field illumination, respectively. (i) shows blood vessel rupture by intravascular tumor cell proliferation and concomitant albumin leaking on day 28. Stereo (c), (d), (g), (h) or confocal (a), (b), (i) fluorescence microscope.

Heterogeneity of BBB integrity

At weeks 1–2, rhodamine-labeled albumin remained within the blood vessel lumen if the tumor cells were inside the vessels, suggesting that BBB was still intact (Fig. 2b). Blood vessels remained patent (Fig. 1b) or became leaky in areas where tumor was growing extravascularly (Fig. 3a and Fig. 1d). This was especially evident in areas adjacent to the tumor core (Figs. 3b and 3c). The distribution of rho-albumin in the periphery was heterogeneous indicating heterogeneity of the BBB in different areas of tumor growth.

Figure 3. Heterogeneity of vascular integrity. MDA-MB-435GFP breast cancer cells were injected into the left cardiac ventricle of female nude mice. Intact brains recovered on days 7 (a) and 28 (b) and (c) from mice injected i.v. with rhodamine–albumin were examined by confocal microscopy. (c) is the same field as (b) recorded with rhodamine filters only. Note that the degree of vascular leak (arrows in (a) and (c) indicate areas containing the most extravascular rhodamine) does not necessarily correspond to tumor cell density (GFP; asterisk in (b) indicates area of highest tumor density). Yellow vasculature in (b) is due to overlapping rhodamine-containing vasculature surrounded with GFP-labeled tumor cells.

Discussion

The view that tumor cells proliferate only after extravasation is discouraging since this makes the diagnosis and treatment of early brain metastasis extremely difficult.9, 10 Although it has been reported that tumor cells affect the permeability of the BBB, most brain tumors and metastasis to the brain are chemo-resistant.11–13 Observation of GFP-labeled tumor cell metastases in isolated perfused lung indicated, however, that hematogenous lung metastasis originate from the proliferation of intravascular tumor cells rather than cells that extravasated the vasculature.14 This observation contrasts traditional views that circulating metastatic cells must extravasate the vasculature before proliferating into colonies.9, 10, 15, 16 This issue is extremely important for brain metastasis, since the formation of extensive intravascular metastasis could provide a therapeutic treatment window despite an intact BBB.

Injection of GFP-labeled tumor cells in vivo allowed us to observe the pathogenesis of brain metastasis with excellent spatial resolution.17 This greatly facilitated a broad investigation into the development of brain metastasis. Our observations provide the first clear evidence that brain metastases are established by two distinct mechanisms—extravasation and intravascular proliferation. Similar to the recent observations of Yamauchi et al.,18 the extravasated cells remained in close proximity to the blood vessels and seemed to preferentially migrate and replicate along the outside of the blood vessel wall. Intravascular growth, on the other hand, progressed from several cells that, at least initially, did not obstruct blood flow. The continued proliferation of these small nodules, however, resulted in blood vessel occlusion, which, upon further growth, led to vessel rupture with concomitant downstream dissemination of tumor cells.

The reasons why some brain metastases begin intravascularly and some extravascularly remain unknown. Conceivably, this might be due to difference in the biological properties of the tumor cells or inherent differences in vascular structure at different locations. The tumor emboli that mediate occlusion of blood vessels together with brain tissue necrosis provides an explanation of why, in some patients, imaging findings are only of a small tumor nidus but with significant vasogenic edema and/or necrosis.19 The constant release of tumor cells to downstream areas from sites of intravascular tumor growth explains, in part, why brain metastasis progress so aggressively.

Our results show a temporal and spatial heterogeneity of the integrity of the BBB in the same brain metastatic setting. Although tumor cells can affect BBB permeability,11–13 our results indicate that there are significant areas of tumor that are protected by the BBB providing, in part, an explanation for the inability of chemotherapeutic regimens to affect brain metastasis. These results might explain the controversy between the lack of BBB integrity and the lack of chemotherapeutic efficacy. Taken together, these results raise the possibility that chemotherapy will not only affect tumor cells not protected by the BBB, but also eradicate chemotherapy-sensitive intravascular tumor growth thereby preventing downstream tumor spread and intravascular occlusion.

Our observations indicate that a large fraction of breast cancer brain metastasis originate from endothelium attached tumor cells that proliferate within the vasculature. These cells can give rise to distant metastasis by flowing downstream within the vasculature, which results in a massive dissemination of tumor cells. As the intravascular colonies grow, they can also destroy the vessel wall and extravasate and migrate along pre-existing blood vessels. In conclusion, our results provide a framework about how breast cancer brain metastases initiate and progress and the heterogeneity of BBB function in brain metastases. These results may provide for a better understand of the clinical presentations of brain metastasis and potentially identify different targets for therapeutic intervention.