Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2020 Sep 2.
Published in final edited form as: Invest Radiol. 2011 Nov;46(11):718–725. doi: 10.1097/RLI.0b013e318226c427

In Vivo Characterization of Changing Blood-Tumor Barrier Permeability in a Mouse Model of Breast Cancer Metastasis

A Complementary Magnetic Resonance Imaging Approach

Dean B Percy *, Emeline J Ribot *, Yuhua Chen *, Catherine McFadden *, Carmen Simedrea †, Patricia S Steeg ‡, Ann F Chambers †, Paula J Foster *
PMCID: PMC7465076  NIHMSID: NIHMS1614113  PMID: 21788908

Abstract

Objectives:

The current lack of efficacy for any chemo- or molecular therapeutic in the treatment of brain metastases is thought to be due, in part, to the heterogeneous permeability of the blood-brain-barrier (BBB). Little is known about how heterogeneous permeability develops, or how it varies among individual metastases. Understanding the BBB’s role in metastasis will be crucial to the development of new, more effective therapies. In this article, we developed the first magnetic resonance imaging-based strategy to detect and measure the volumes of BBB permeable and nonpermeable metastases and studied the development of altered BBB permeability in metastases in vivo, over time in a mouse model of breast cancer metastasis to the brain.

Materials and Methods:

Animals bearing human experimental brain metastases of breast cancer (231-BR cells) were imaged, using 3-dimensional balanced steady-state free precession to visualize total metastases, and contrast-enhanced T1-weighted spin echo with gadopentetic acid (Gd-DTPA) to visualize which of these displayed contrast enhancement, as Gd-DTPA leakage is indicative of altered BBB permeability.

Results:

Metastases detected 20 days after injection showed no Gd-DTPA enhancement. At day 25, 6.1% ± 6.3% (mean ± standard deviation) of metastases enhanced, and by day 30, 28.1% ± 14.2% enhanced (P < 0.05). Enhancing metastases (mid: 0.14 ± 0.18 mm3, late: 0.24 ± 0.32 mm3) had larger volumes than nonenhancing (mid: 0.04 ± 0.04 mm3, late: 0.09 ± 0.09 mm3, P < 0.05); however, there was no significant difference between the growth rates of the 2.

Conclusions:

A significant number of brain metastases were uniformly nonpermeable, which highlights the need for developing treatment strategies that can overcome the permeability of the BBB. The model developed herein can provide the basis for in vivo evaluation of both BBB permeable and nonpermeable metastases response to therapy.

Keywords: MRI, brain metastasis, blood-brain barrier, breast cancer, permeability, gadolinium

The permeability of the blood-brain-barrier (BBB) plays a crucial role in brain metastasis. This permeability has been implicated in the initial stages of cancer cell extravasation, metastasis growth, vascular remodeling, and response to treatment.1–5 Under normal conditions, the BBB tightly controls the diffusion of substrates in the blood into the surrounding parenchyma.6–8 However, in the presence of a tumor, the local BBB, sometimes referred to as the blood-tumor-barrier (BTB), can become altered. Previous work has shown that the BTB can exhibit heterogeneous permeability to marker proteins or drugs.1,3,9 The population of brain metastases that exhibit little to no BTB permeability are of great clinical interest, as these metastases respond poorly to treatment.2

Gadolinium (Gd)-enhanced contrast magnetic resonance imaging (MRI) is widely regarded as the most accurate diagnostic imaging modality for brain tumors.10 Gd-based contrast agents (which appear hyperintense in T1-weighted [T1w] MR images), cannot cross an intact BBB, and thus their appearance in brain tumors is an indication of altered BTB permeability.11 It is known that tumor detection by contrast-enhanced MRI can be affected by contrast agent type,12,13 concentration,10,14,15 and BTB permeability.16 Furthermore, the clinical need to monitor the status of brain tumor and normal tissue permeability, especially in response to treatment, has been highlighted.17,18 Thus, the ability to visualize the relative numbers of BBB permeable and nonpermeable metastases, and how these numbers change over time, would help shed light not only on what percentage of metastases may not respond to therapy, but also how metastasis detection and permeability may change over time.

Although differential BTB permeability in primary cerebral malignancies has been studied using MRI,19,20 characterizing BTB permeability among metastases presents a unique challenge since there are often multiple tumors at different locations within the brain,7 and it is necessary to sample a large amount of the brain volume. Furthermore, there are relatively few efficient preclinical models available to study the development of cerebral metastases. Yoneda et al21 have developed a highly efficient experimental breast cancer brain metastasis model in mice (human MDA-MB-231-BR cells), using intracardiac (IC) injection, which delivers cells systemically via the arterial system. This model has been used extensively to study experimental brain metastases in mice3,22–24; however, studies on altered BTB permeability have so far been limited to single time point in situ histologic analyses.

Song et al have previously demonstrated that MDA-MB-231-BR metastases can exhibit Gd-enhancement in an experimental metastasis model in rats25; however, there are currently no studies that examine the relative numbers of permeable and nonpermeable brain metastases in vivo, and how these numbers change over time. Furthermore, longitudinal visualization of both BTB permeable and nonpermeable metastases would allow for in vivo monitoring of tumor response to therapy, especially those that attempt to overcome heterogeneous BTB permeability.9 Due to its noninvasive nature and high resolution, MRI provides an excellent basis for studying brain metastasis growth and development in vivo.

Balanced steady-state free precession (bSSFP) is an MRI technique that can achieve a high signal-to-noise ratio even at high spatial resolutions.26 The resulting image also possesses contrast properties of interest to brain tumor detection, as fluids and tumor-associated edema appear hyperintense to surrounding brain tissue.27 bSSFP can be used in 3 dimensions (3D), which allows for visualization of an entire organ volume, with reasonable scan times. As such, it is a strong candidate for detecting cerebral metastases. Heyn et al first demonstrated its use for detecting single cancer cells in vivo in the mouse brain23 and for monitoring the development of breast cancer metastasis in the mouse brain at 1.5 Tesla (T). Bernas et al found bSSFP to be superior to Gd-enhanced T1w 3D-GE in the detection of high-grade gliomas in mice, even without the use of a contrast agent.27 Miraux et al have investigated the use of bSSFP for the visualization of glioma in mouse brain.28 Recently, bSSFP has been used clinically to enhance the differentiation of small pineal cysts and tumors, and to better delineate fine brain structures around tumors for surgical planning.29,30

In this study, there were 3 objectives. First, to determine which MRI pulse sequence (3D bSSFP, 3D T1w gradient echo, and 2D T1w spin echo [SE]) gave the best detection of metastases without the use of contrast agent, to best visualize nonpermeable metastases. Second, to determine which MRI pulse sequence gave the best detection of gadopentetic acid (Gd-DTPA) enhancement in metastases postcontrast, to visualize BBB permeability. With these results in mind, the third goal was to examine the onset and extent of BBB permeability in vivo over time in an experimental mouse model of breast cancer metastasis to the brain, using the sequences selected from the previous objectives. This allowed for the quantification of the relative numbers and volumes of permeable and nonpermeable metastases in the whole brain throughout the experiment.

MATERIALS AND METHODS

Cell Labeling

MDA-MB-231-BR human breast cancer cells, which are preferentially brain metastatic,21 were labeled with micron-sized paramagnetic iron oxide (Bangs Laboratories, US), and transfected with enhanced green fluorescent protein to allow for detection by fluorescence microscopy, as described previously.23 Cell labeling was assessed as has been previously described, and had no significant effect on cell viability.31,32

Animal Model

All protocols for these experiments were approved by our institution’s Animal Care Committee in accordance with the policies established by the Canadian Council on Animal Care. An intracardiac (IC) injection of a 0.1 mL suspension of 1.75 × 105 MPIO-labeled MDA-MB-231-BR cells was performed in female nu/nu mice aged 6 to 8 weeks (Charles River Laboratories, US). Animals were anesthetized using 2% isoflurane gas mixed with 100% O2. A 29-gauge needle was inserted under the sternum into the left ventricle of the beating heart to inject the cell suspension.2,23

In vivo MR imaging

All imaging experiments were performed using a 3T GE clinical scanner with a custom built high performance insert gradient coil (maximum gradient strength = 500 mT/m, and peak slew rate = 3000 T/m/s) and a custom solenoid radio-frequency mouse head coil (inner diameter of 1.5 cm).27 For imaging, animals were anesthetized using 2% isoflurane gas in 100% O2, and body temperature was maintained through convectional heating by warm saline bags. All mice were imaged 1 day post IC injection to determine injection efficiency, using 3D bSSFP (resolution: 150 × 150 × 200 μm, repetition time/echo time (TR/TE) = 8/4, flip angle = 35°, scan time 10 minutes). Images were viewed and analyzed using OsiriX image software (open source). All statistics were performed with a 1-way, nonparametric analysis of variance (ANOVA), or nonparametric t test using GraphPad Prism software. Only animals that presented with MR signal voids at day 1 (indicating successful IC injection) were included in the study.

MR Pulse Sequences

To address the first 2 objectives, 3 clinically relevant MR pulse sequences were compared for the ability to detect brain metastases: (i) without the use of Gd contrast enhancement (which would include both BBB permeable and nonpermeable metastases, and thus total metastatic burden), and (ii) with the use Gd-DTPA to determine which of these metastases exhibit contrast enhancement (indicative of altered BBB permeability). Enhancing metastases were defined as the one which appeared hyperintense to surrounding brain tissue in T1w spoiled gradient recovered (SPGR), T1w SE, and bSSFP after administration of Gd-DTPA, as judged by the observer comparing pre- and post-Gad images. The same 4 mice were imaged on consecutive days, 3 to 4 weeks post IC injection of cells using 3D bSSFP (resolution: 100 × 100 × 200 μm, TR/TE = 8/4 milliseconds, flip angle = 35°, scan time 40 minutes), 3D T1w SPGR (resolution: 100 × 100 × 200 μm, TR/TE = 15/3 milliseconds, flip angle = 30°, scan time 43 minutes), and 2D T1w SE (resolution: 156 × 156 × 500 μm, TR/TE = 600/20 milliseconds, number of slices = 17, scan time 15 minutes), respectively. Images were acquired pre- and post-intraperitoneal (IP) injection of 0.2 mL Gd-DTPA (Magnevist, Schering, US). As the T1wSE sequence had a shorter scan time, multiple consecutive scans (4–6) were performed, and timing was adjusted so that SE images were acquired at the same approximate time interval postcontrast injection as bSSFP and SPGR. The numbers of metastases detected with each sequence were obtained by a blind count of each image data set by a single observer, both pre- and postcontrast. Because the T1wSE sequence had a smaller field of view compared with both bSSFP and SPGR, metastases not present in the field of view of all 3 sequences were not counted. The ability to detect metastases in noncontrast bSSFP images was also compared with 2D T2w SE images (resolution: 117 × 117 × 400 μm, TR/TE = 2400/80 milliseconds, number of slices = 27, scan time: 123 minutes), since this is another very common approach for tumor visualization in preclinical and clinical studies. Mice were imaged with bSSFP on day 30 and T2wSE on day 31 (N = 2).

Visualization of Changing BBB Permeability Over Time

Based on the results of our pulse sequence experiments, 7 mice were injected (as per above) and imaged using the 2D T1wSE sequence (resolution: 117 × 117 × 400 μm, TR/TE = 600/20, flip angle = 35° scan time 15 minutes) 60 minutes post IP injection of 0.2 mL of Magnevist for brain imaging on days 19, 24, and 29 post IC injection of cells to visualize enhancing metastases. Optimal timing and dosage of IP Gd-DTPA were based on the literature33 and our own unpublished data. The same animals were imaged again the next day, days 20, 25, and 30, using the noncontrast 3D bSSFP sequence (resolution: 100 × 100 × 200 μm, TR/TE = 16/8, flip angle = 35°, scan time: 30 minutes) to obtain total metastatic burden in the brain. One animal was killed at each time point, to provide corresponding in situ permeability data (see below). Together, these constituted early (days 19/20, N = 6), mid (days 24/25, N = 5), and late (days 29/30, N = 4) time points. An enhancing metastasis was defined as the one which appeared hyperintense to surrounding brain tissue in T1w SE following Gd-DTPA injection. A nonenhancing metastasis was defined as the one that was detected on noncontrast bSSFP images, but was not hyperintense on corresponding T1wSE images after administration of Gd-DTPA. Metastases were counted as above, and an enhancing fraction was calculated by dividing the number of permeable metastases (visible by Gd-enhanced T1wSE) by the total number of metastases (visible by noncontrast bSSFP) in each mouse. Volumes were calculated by a single observer, using manual hand-tracing of the tumor boundary in each slice from the 3D bSSFP images, and a volume calculation algorithm in OsiriX. Window levels were adjusted to provide optimum contrast between metastases and the surrounding tissue; however for consistency, levels were kept constant when tracing individual metastases. For volume measurements, only mice that survived until end point were analyzed (N = 4).

In Situ Permeability Assay

An in situ permeability assay using fluorescent dextran was performed on mice at each imaging time point to compare with in vivo Gd-enhancement observed by MRI.3 Animals were anesthetized using 2% isoflurane gas in 100% O2 and injected intravenously with 1.5 mg of 3 kDa lysine fixable Texas Red Dextran (Invitrogen, US) in 0.15 mL saline. After 20 minutes of dextran injection, mice were injected IC with heparin (Pharmaceutical Partners of Canada, Canada) and perfusion fixed with saline and 4% formalin, a modified protocol from Choi et al.34 Brains were excised and placed in descending concentrations of sucrose (30%, 20%, 10%), cryopreserved, mounted in optimal cutting temperature compound and cut into 20 μm sections. Dextran leakage was visualized using red fluorescence (530/585 nm excitation, 615 nm emission) and MDA-MB-231-BR cells were visualized using green fluorescence (470 nm excitation, 525 nm emission) microscopy on an Axio Imager.A1 microscope (Zeiss, Germany). Images were acquired using a Retiga EXi (QImaging, Canada) digital camera. Selected sections were then stained with hematoxylin and eosin,35 to provide an anatomic reference to the fluorescent images, and imaged using the same microscope/camera setup as above.

RESULTS

MR Pulse Sequence Selection

In experiment 1, our 2 goals were to determine which sequences would provide the best detection of (i) non–Gd-enhancing (and thus nonpermeable) metastases, and (ii) Gd-enhancing (and thus permeable) metastases. We compared the metastasis detection both before and after injection of Gd-DTPA contrast agent for each sequence. Representative images are shown in Figure 1. All metastases detected using noncontrast bSSFP (Fig. 1C) appeared as signal hyperintensities (mean ± standard deviation) (30.0 ± 9.4, 120 total, 4 mice). Metastases detected in noncontrast SPGR (Fig. 1A) appeared as signal hypointensities (18.3 ± 5.6, 73 total, 4 mice). Very few metastases were detected using noncontrast T1wSE (Fig. 1B), and these were also detected as signal hypointensities (1.3 ± 1.0, 5 total, 4 mice). In noncontrast images, the number of metastases detected using bSSFP or SPGR was significantly greater than with T1wSE.

FIGURE 1.

FIGURE 1.

Open in a new tab

Comparison of metastasis detection pre- and post Gd-DTPA. Axial MR images of the mouse brain, with arrows indicating metastases. A to C show images of each sequence, before injection of Gd-DTPA contrast agent. Metastases appeared as signal hypointensities in noncontrast SPGR (A). Very few metastases were detected in noncontrast T1wSE, and appeared hypointense (C). Metastases appeared hyperintense in noncontrast bSSFP images (C). D to F show the corresponding images, post Gd-DTPA injection. Metastases appeared either hyperintense or isointense in postcontrast SPGR (D). Metastases were hyperintense in postcontrast T1wSE (E) and bSSFP (F) images. In noncontrast images, metastases are easiest to visualize with bSSFP. However, the only significant (P < 0.05) differences in metastasis detection were seen in noncontrast T1wSE and postcontrast SPGR (G).

Interestingly, in postcontrast bSSFP images (Fig. 1F), metastases still appeared hyperintense (29.0 ± 10.9, 116 total, 4 mice), and thus there was no significant difference in the number of metastases detected by bSSFP pre- and postcontrast. Surprisingly, the number of metastases detected in postcontrast SPGR (2.5 ± 0.6, 10 total, 4 mice) images (Fig. 1D) was significantly decreased compared with precontrast images (P < 0.05). As expected, many more metastases were detected in postcontrast T1wSE (25.8 ± 11.1, 103 total, 4 mice) images (Fig. 1E) compared with precontrast (P < 0.05). Figure 1G summarizes the results of the metastasis detection for each sequence pre- and post-Gd-DTPA injection. For our purposes, bSSFP provided superior metastasis detection without the use of contrast agent, and thus was the sequence of choice for imaging nonpermeable (vis-à-vis, total) metastases. T1wSE with Gd-DTPA was chosen to distinguish which of the metastases exhibited contrast enhancement, and thus BTB permeability.

As well, since noncontrast T2w imaging is commonly used for MRI of primary brain tumors, we also compared noncontrast bSSFP and T2wSE. Representative images for each sequence from the same animal are shown in Figure A1, online only, supplemental digital content, available at: http://links.lww.com/RLI/A48. The T2wSE (Fig. A1A, online only, supplemental digital content, available at: http://links.lww.com/RLI/A48) and bSSFP (Fig. A1B, online only, supplemental digital content, available at: http://links.lww.com/RLI/A48) sequences produced similar tissue contrast and both exhibited similar metastasis detection (Fig. A1C, online only, supplemental digital content, available at: http://links.lww.com/RLI/A48).

In Vivo BBB Permeability Over Time

On the basis of our results from experiment 1, in experiment 2 we imaged each mouse using T1wSE with Gd-DTPA on days 19, 24, and 29 and then with noncontrast bSSFP on days 20, 25, and 30 post cell injection. These time points were referred to as early (days 19/20), mid (days 24/25), and late (days 29/30). Representative images from the same mouse over time are shown in Figure 2. At the early time point, no enhancing metastases were visualized using Gd-enhanced T1wSE (Fig. 2A), which suggests that the BBB is not permeable at this early time point in our model. There were, however, metastases detected using bSSFP at this early time point (Fig. 2D, arrow). Enhancing metastases were detected using Gd-enhanced T1wSE at mid (2.2 ± 2.5, 11 total, 5 mice) and late (8.8 ± 3.6, 36 total, 4 mice) time points (Figs. 2B, C, large arrows). The number of metastases detected with bSSFP increased significantly between the early (10.5 ± 3.4, 63 total, 6 mice) and mid (34.0 ± 9.8, 170 total, 5 mice) time points (Fig. 2E, arrows), but was not different between the mid and late (33.8 ± 9.2, 140 total, 4 mice) time points (Fig. 2F, arrows). There was a significant difference between the number of metastases detected in bSSFP images and the number of enhancing metastases detected in T1wSE images at the mid and late time points (P < 0.01) (Fig. 2G). The fraction of the total metastases that exhibited Gd-enhancement at each time point are shown in Figure 3H. The fraction of enhancing metastases at late time point (28.0% ± 14.2%) was significantly greater than that at mid time point (6.1% ± 6.3%, P < 0.05). Overall, 3 distinct populations of metastases were observed: (i) those that were visible in bSSFP images and enhanced with Gd in T1wSE at the same time point; (ii) those that were detected with bSSFP first and then enhanced at a later time point, and (iii) those that were detectable with bSSFP but did not enhance with Gd through to end point. Most of the enhancing metastases detected at mid time point were of the first subset (75%). However, the majority of enhancing metastases that enhanced at late time point were of the second subset (72%). It was also observed that once a metastasis enhanced with Gd, it continued to exhibit Gd enhancement at later time points.

FIGURE 2.

FIGURE 2.

Open in a new tab

In vivo visualization of altered BBB permeability in the same animal over time. Axial MR images of the mouse brain. A to C show Gd-enhanced T1wSE, at each of early, mid, and late time points. D to F show the corresponding noncontrast bSSFP image. A metastasis is visible at early time point with bSSFP (D, arrow), but does not exhibit Gd-enhancement (A). At mid time point, this metastasis exhibited Gd-enhancement (B, large arrow). There are also several smaller metastases visible with bSSFP at mid time point (E, arrows) that do not enhance. At late time point, the large metastasis exhibited extensive Gd-enhancement (C, large arrow); however, the small metastases visible with bSSFP (F, arrows) still did not enhance (F). There were a small number of metastases visible with bSSFP at early time point; however, none of these exhibited Gd-enhancement. There was a significant difference between the number of metastases detected with bSSFP and the number detected with Gd-enhanced T1wSE at mid and late time points (P < 0.01) (G). When taken as a fraction of total metastases detected (H), there was a significant difference between the number of enhancing metastases at mid and late time points (P < 0.05). The ventricles can also be visualized (*) in these images.

FIGURE 3.

FIGURE 3.

Open in a new tab

Volume measurements of enhancing and nonenhancing metastases. A, At both mid and late time points, the average volume of enhancing metastases was significantly larger than nonenhancing metastases (P < 0.05). However, there was a wide range of volumes for both enhancing and nonenhancing metastases (B). There also appeared to be a minimum volume threshold of enhancing metastases, although being larger than this did not guarantee enhancement. C, 3D volume rendering of a mouse brain in the coronal plane, from the same mouse at each time point. Gd-enhancing metastases are rendered in red and nonenhancing metastases are shown in green. Neither volume nor position in the brain appear to have an influence on whether a metastasis enhanced or not.

The results of metastasis volume measurements for all animals that survived until end point are shown in Figure 3. The mean volume of nonenhancing metastases at the early time point was 0.01 ± 0.01 mm3 (mean ± standard deviation, 38 total, 4 mice). There were no enhancing metastases at the early time point. The mean volume of enhancing metastases, at mid (0.14 ± 0.18 mm3, 8 total, 4 mice) and late (0.24 ± 0.32 mm3, 46 total, 4 mice) time points, was significantly greater than that of the nonenhancing metastases (mid: 0.04 ± 0.04 mm3, 117 total, 4 mice; late: 0.09 ± 0.09 mm3, 104 total, 4 mice, P < 0.05) as shown in Figure 3A. Interestingly, although the volumes of nonenhancing metastases were significantly different at each time point (P < 0.05), there was no significant difference in the average volumes of enhancing metastases between mid and late time points. There was also a wide range of volumes observed (Fig. 3B), even among enhancing and nonenhancing metastases. There appears to be a threshold below which no Gd-enhancing metastases were observed, the smallest being 0.02 mm3 (Fig. 3B). The growth rates of metastases between the mid and late time points were determined by calculating the volume change between the 2 time points (normalized by the volume at the earlier of the 2 time points). This was not calculated for early and mid time points because there were too few enhancing metastases at the early time point. Interestingly, there was no significant difference in the growth rates of nonenhancing metastases, metastases that first enhanced at the mid time point or metastases that first enhanced at the late time point (not shown).

Figure 3C shows a 3D rendering of the metastases in a mouse brain in the coronal view from the same mouse at early, mid, and late time points. Metastases that exhibited Gd-enhancement are shown in red, and those that were detectable with bSSFP but did not exhibit Gd-enhancement are shown in green. Neither metastasis volume nor location in the brain appear to have any bearing on whether a metastasis displayed Gd-enhancement or not, as both enhancing and nonenhancing metastases were detected throughout the brain. Although there was a trend for enhancing metastases to be larger, this was not always the case: there were also many large metastases that did not exhibit Gd-enhancement (Figs. 4A, C, arrow) and small metastases that did exhibit Gd-enhancement (Figs. 4B, D, arrows).

FIGURE 4.

FIGURE 4.

Open in a new tab

Examples of volume and permeability heterogeneity in metastases. Axial MR images of the mouse brain. A, B show Gd-enhanced T1wSE images from 2 different mice. C, D show the corresponding noncontrast bSSFP images. A large metastasis (0.42 mm3) near the lateral ventricle was visible with bSSFP (C, arrow), but did not exhibit Gd-enhancement (A), while a nearby metastasis visible with bSSFP (C) did exhibit Gd-enhancement (A, arrow). Two very small metastases (left: 0.04 mm3, right: 0.06 mm3) in the frontal cerebral cortex of another animal were visible with bSSFP (D, arrows) and exhibited Gd-enhancement (B, arrows).

In Situ Permeability

Representative Texas Red dextran fluorescence images are shown with corresponding bSSFP and Gd-enhanced T1wSE images in Figures 5 and 6. bSSFP (Fig. 5A) and T1wSE (Fig. 5B) images reveal a large, Gd-enhancing metastasis at mid time point. Fluorescent microscopy of the same metastasis (Fig. 5C) reveals extensive dextran leakage (red fluorescence) in the surrounding brain parenchyma and tumor cells (green fluorescence). The same section H&E stained is shown in Figure 5D. Figure 6 shows a small, nonenhancing metastasis at the early time point. The metastasis was detectable with bSSFP (Fig. 6A), but not Gd-enhanced T1wSE (Fig. 6B). Fluorescent microscopy revealed no dextran within the lesion (Fig. 6C, green fluorescence); however, dextran can be seen around the vessels just above the metastasis (red fluorescence). The same section stained with H&E is shown in Figure 6D. Dextran data agreed with MR data in most cases; however, there were some instances where Gd-enhanced MRI revealed BBB permeability and dextran did not. This disagreement was often detected in the larger, more diffuse metastases, as indicated by H&E histology where the dextran may have collected in fluid-filled cavities, but ultimately was lost during ex vivo tissue processing. Based on histologic images, following 2 types of metastases were seen: large, diffuse mostly fluid-filled lesions, as in Figure 6, and more compact, tightly packed clusters of cells, as in Figure 5. None of these types were found to exhibit preferential BBB permeability by either dextran or Gd-DTPA.

FIGURE 5.

FIGURE 5.

Open in a new tab

BBB permeability with Gd-enhanced MRI confirmed with in situ fluorescence. Axial MR images of the mouse brain reveal a metastasis visible with bSSFP (A, arrow) that exhibited Gd-enhancement with T1wSE (B, arrow) at mid time point. C, The same metastasis visualized with fluorescent microscopy, appeared as 2 separate clusters of MDA-231-BR cells (green fluorescence), and exhibited extensive fluorescent dextran permeability in the surrounding parenchyma (red fluorescence). D, The same section stained with H&E, to provide anatomic reference, shows a compact tumor connected to a larger, more diffuse tumor cell cluster.

FIGURE 6.

FIGURE 6.

Open in a new tab

Non-BBB permeable metastasis detected with MRI confirmed with in situ fluorescence. Axial MR images of the mouse brain reveal a small metastasis that was visible with bSSFP (A, arrow) that did not exhibit Gd-enhancement with T1wSE (B) at early time point. C, The same metastasis visualized with fluorescent microscopy appears as a diffuse collection of several MDA-231-BR cells (green fluorescence) with no fluorescent dextran leakage visible in the surrounding tissue. There is, however, dextran leakage surrounding vessels just above the tumor (red fluorescence). D, The same section stained with H&E reveals large lesions of empty space surrounding the cancer cells.

DISCUSSION

The permeability of the BBB and BTB plays a very important role in brain metastasis.1–5,36 The initial arrest and extravasation of cancer cells, the recruitment of new vasculature, and the effective delivery of chemotherapeutics are all influenced by BBB permeability.5 The need to understand the nature of this altered permeability has been highlighted2,3; however, little is known about the development of BTB permeability of metastases. A wide range of cellular and biochemical factors can influence BBB permeability; however, tumors in the brain often have underdeveloped vasculature which can lead to leaky vessels, and lack of a functional BBB.5,37 BTB permeability to conventional contrast agents such as Gd-DTPA is due to fenestrations in capillaries, and dependent on molecule size.18 Fenestrae can be caused by vascular changes, such as angiogenesis and vascular remodeling,3 and by changes in the surrounding microenvironment.13,37 For example, increased levels of extravascular fibrin (which forms as a result of increased angiogenesis and vessel permeability) have been detected in vivo using Gd-enhanced MRI.13 Since BTB permeability must play a role in the “metastatic cascade,”38 it is no doubt itself a dynamic process. Thus, techniques that allow for longitudinal, in vivo detection and monitoring of BBB and BTB permeability would prove essential in studying the development of altered BBB in metastases. In this article, we present a novel, MR-based approach to longitudinally and noninvasively observe heterogeneous Gd-enhancement among developing metastases in the entire mouse brain. Using high resolution, 3D bSSFP, in tandem with conventional T1wSE with Gd-DTPA, we were able to detect and quantify the relative numbers of Gd-enhancing and -nonenhancing metastases in vivo, over time in an experimental mouse model of MDA-MB-231-BR metastases to the brain. Since Gd-DTPA cannot cross an intact BBB, Gd-enhancement can serve as a functional in vivo marker for BBB permeability.39 To our knowledge, we are the first to use MRI to examine the development of altered BTB permeability in metastases in vivo, noninvasively over time.

The initial objectives of this study were to determine which MRI pulse sequences had the best (i) metastasis detection without Gd-DTPA and (ii) visualization of Gd-DTPA uptake. bSSFP was found to have greatest sensitivity to metastases in precontrast images. T1wSE was chosen as the sequence of choice for detecting Gd-DTPA uptake postcontrast injection for 2 reasons. First, postcontrast T1wSE images had a higher sensitivity to metastases than postcontrast T1w SPGR. Second, T1wSE had a higher specificity to Gd-DTPA enhancement than bSSFP, since metastases appeared hyperintense in both pre- and postcontrast bSSFP, whereas they only appeared hyperintense in postcontrast T1wSE. Thus, using these 2 sequences in tandem allowed for visualization and quantification of both BTB permeable and nonpermeable metastases.

The observation that early, small experimental brain metastases are uniformly nonpermeable highlights the need for the development of BBB-permeable drugs for effective prevention, especially given that early treatment of brain metastases often leads to better patient prognosis.7 Most standard chemotherapeutic drugs are BBB impermeable.3,9 There are drugs that display BBB permeability, such as vorinostat40 and temozolomide,41 however none of these drugs have activity in the metastatic breast cancer setting (although the preventive activity of vorinostat has been suggested in preclinical models).40 There are also techniques that attempt to increase the permeability of the BBB itself. Focused ultrasound-induced BBB permeability has been shown to increase chemotherapeutic delivery in a rat model of glioblastoma.42 As well, Hu et al demonstrated an increased delivery of herceptin to brain metastases in several different cell lines (including breast cancer) by concomitant administration of phosphodiesterase-5 inhibitors.43 In vivo assessment of such novel strategies is crucial, and we believe the ability to track both permeable and nonpermeable metastasis volumes presented herein can help provide longitudinal assessment of treatment efficacy.

Since Gd-enhanced contrast MRI is widely recognized as the most accurate diagnostic tool in the detection of brain metastases,10 our results point at a large fraction of metastases that would not be optimally imaged on a Gd-enhanced MRI scan, due to the large proportion of metastases that may not have achieved BBB permeability. Our findings also suggest that this fraction would be larger at early time points, when there is significantly less BBB permeability. This has clinical implications, as early detection and treatment of brain metastases often leads to better management.7 Although bSSFP is not a prominent clinical MRI sequence in the diagnosis of brain metastases, its utility has been demonstrated in improving diagnosis and treatment planning of brain tumors.29,30,44 Notably, Notle et al demonstrated the superior sensitivity of bSSFP in the detection of pineal cysts, when compared with other conventional MRI pulse sequences, which approached that of autoptic studies.44 Due to its high resolution and ability to detect non-BBB permeable metastases, we believe further investigation into the use of bSSFP in human brain metastasis detection is warranted, as it may prove to be a valuable tool in the diagnosis of brain metastases, especially at the early stages of systemic metastasis, when the BTB may not be permeable.

Others have studied the relationship between brain metastases and permeability using in situ techniques. Lockman et al showed that passive permeability to Texas red dextran was unrelated to the size of the metastases or the morphology (compact or diffuse) in mice inoculated with the 231-BR-Her2 cell line.3 On the other hand, a histologic examination of 8 human tumor cell lines, performed by Zhang et al, showed 2 patterns of metastasis growth.1 Small, isolated metastases in the parenchyma of the brain, described as nodules, showed no permeability. Diffuse metastases were not permeable until they coalesced to form large masses. These results suggest that the permeability of the BBB is related to the growth pattern/morphology and size of the lesions. Together with our results, these studies reveal a complex relationship between metastasis size and permeability, with size being a potentially necessary, but not sufficient condition for BTB permeability.

Furthermore, in vivo growth measurements of our study suggest that not only is metastasis volume not related to BBB permeability, but that growth rate is similarly unaffected by BTB permeability. Our results do reveal, however, that BTB permeable metastases reached a maximum volume somewhere between the mid and late time points, whereas nonpermeable metastases continued to increase in volume during the same time period. Based on this, and the fact that most enhancing metastases at late time point were already present as nonenhancing metastases at earlier time points, timing may play a crucial role in metastasis permeability. This would suggest a potential series of conditions that must be met by a metastasis to achieve BTB permeability.

There is an important caveat concerning MRI measured metastasis volumes. Since fluids appear hyperintense in bSSFP (and in T2w) images, volume measurements of metastases using bSSFP include tumor-associated edema, which can lead to misrepresentation of actual tumor size.45 Since BBB permeability can lead to increased edema, this may lead to overestimation of the volume of permeable metastases. It is possible that this may also contribute to the significant volume differences between permeable and nonpermeable metastases, as measured in this study.

In the future, our technique could be used with more frequent temporal sampling, which may provide further insight into the relationship between metastasis size, growth rate, and BBB permeability. As well, detailed immunohistochemical and histologic analyses can be planned using in vivo MRI enhancement data, to examine potential biomarkers involved in BBB and BTB permeability, which would further relate the extensive in situ data to in vivo dynamics. By being able to track both BBB permeable and nonpermeable metastases in the same animal over time, our approach can also shed light on some of the other necessary or sufficient conditions for metastases to have altered BBB permeability, such as location within the brain, temporal information, relation to blood flow, and even exposure to radiation, all factors that have been discussed in depth with regards to BBB permeability and brain metastases.4

In summary, our in vivo MRI approach allows for much needed study of the dynamic nature of altered BBB permeability in brain metastases. Our findings have both preclinical and clinical significance in metastasis detection and treatment. From a preclinical standpoint, the understanding of altered BTB permeability can be improved using in vivo imaging techniques that allow for longitudinal detection and quantification of permeable and nonpermeable metastases. For example, we have shown that a large fraction of metastases are impermeable to Gd-DTPA at early stages of brain metastasis, and this fraction decreases significantly over time. We have also shown that BBB permeable metastases are significantly larger than non-BBB permeable metastases; however, size alone is not a sufficient condition to predict BBB permeability, and there was found to be no difference in the growth rates between BBB permeable and non permeable metastases.

The clinical significance is 2-fold. First, many metastases detected did not exhibit Gd-DTPA enhancement, and were only detectable using bSSFP. This warrants further study of bSSFP and its ability to detect small, nonpermeable metastases and a comparison of bSSFP to the conventional MRI pulse sequences in a clinical setting, to evaluate the possibility of improving early detection of brain metastases. The translational significance is the ability to use this model to compare and contrast the in vivo response of BBB permeable and nonpermeable brain metastases to chemotherapeutics, both conventional and those targeted to cross the BBB. Such an assessment would assist in the development of more effective chemotherapeutic strategies that would invariably lead to better clinical management of patients with brain metastases.

Supplementary Material

Supplemental data

Acknowledgments

Supported by the US Department of Defense Breast Cancer Research Program (grant W81XWH-06–2–0033 to PSS, AFC, and PJF). Funded by the Canadian Institute of Health Research and the University of Western Ontario (to D.B.P.). Ann F. Chambers is Canada Research Chair in Oncology and receives salary support from the Canada Research Chairs Program. Funded by the Intramural research program of the NCI (to P.S.S.).

Footnotes

Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s Web site (www.investigativeradiology.com).

REFERENCES

  • 1.Zhang RD, Price JE, Fujimaki T, et al. Differential permeability of the blood-brain barrier in experimental brain metastases produced by human neoplasms implanted into nude mice. Am J Pathol. 1992;141:1115–1124. [PMC free article] [PubMed] [Google Scholar]
  • 2.Palmieri D, Smith QR, Lockman PR, et al. Brain metastases of breast cancer. Breast Dis. 2006;26:139–147. [DOI] [PubMed] [Google Scholar]
  • 3.Lockman PR, Mittapalli RK, Taskar KS, et al. Heterogeneous blood-tumor barrier permeability determines drug efficacy in experimental brain metastases of breast cancer. Clin Cancer Res. 2010;16:5664–5678. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Fidler IJ. The role of the organ microenvironment in brain metastasis. Semin Cancer Biol. 2011;21:107–112. [DOI] [PubMed] [Google Scholar]
  • 5.Arshad F, Wang L, Sy C, et al. Blood-brain barrier integrity and breast cancer metastasis to the brain. Patholog Res Int. 2010;2011:920509. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Strugar J, Rothbart D, Harrington W, et al. Vascular permeability factor in brain metastases: correlation with vasogenic brain edema and tumor angiogenesis. J Neurosurg. 1994;81:560–566. [DOI] [PubMed] [Google Scholar]
  • 7.Klos KJ, O’Neill BP. Brain metastases. Neurologist. 2004;10:31–46. [DOI] [PubMed] [Google Scholar]
  • 8.Engelhardt B, Sorokin L. The blood-brain and the blood-cerebrospinal fluid barriers: function and dysfunction. Semin Immunopathol. 2009;31:497–511. [DOI] [PubMed] [Google Scholar]
  • 9.Thomas FC, Taskar K, Rudraraju V, et al. Uptake of ANG1005, a novel paclitaxel derivative, through the blood-brain barrier into brain and experimental brain metastases of breast cancer. Pharm Res. 2009;26:2486–2494. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Anzalone N, Gerevini S, Scotti R, et al. Detection of cerebral metastases on magnetic resonance imaging: intraindividual comparison of gadobutrol with gadopentetate dimeglumine. Acta Radiol. 2009;50:933–940. [DOI] [PubMed] [Google Scholar]
  • 11.Brant-Zawadzki M, Berry I, Osaki L, et al. Gd-DTPA in clinical MR of the brain: 1. Intraaxial lesions. Am J Roentgenol. 1986;147:1223–1230. [DOI] [PubMed] [Google Scholar]
  • 12.Katakami N, Inaba Y, Sugata S, et al. Magnetic resonance evaluation of brain metastases from systemic malignances with two doses of gadobutrol 1.0 M compared with gadoteridol: a multicenter, phase II/III Study in patients with known or suspected brain metastases. Invest Radiol. 2011;46:411–418. [DOI] [PubMed] [Google Scholar]
  • 13.Morelli JN, Runge VM, Williams JM, et al. Evaluation of a fibrin-binding gadolinium chelate peptide tetramer in a brain glioma model. Invest Radiol. 2011;46:169–177. [DOI] [PubMed] [Google Scholar]
  • 14.Attenberger UI, Runge VM Jackson CB, et al. Comparative evaluation of lesion enhancement using 1 M gadobutrol vs. 2 conventional gadolinium chelates, all at a dose of 0.1 mmol/kg, in a rat brain tumor model at 3 T. Invest Radiol. 2009;44:251–256. [DOI] [PubMed] [Google Scholar]
  • 15.Giesel FL, Mehndiratta A, Risse F, et al. Intraindividual comparison between gadopentetate dimeglumine and gadobutrol for magnetic resonance perfusion in normal brain and intracranial tumors at 3 Tesla. Acta Radiol. 2009;50: 521–530. [DOI] [PubMed] [Google Scholar]
  • 16.Sarin H, Kanevsky AS, Wu H, et al. Effective transvascular delivery of nanoparticles across the blood-brain tumor barrier into malignant glioma cells. J Transl Med. 2008;6:80. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Sorensen AG. Science to practice: blood-brain barrier leakage–one size does not fit all. Radiology. 2010;257:303–304. [DOI] [PubMed] [Google Scholar]
  • 18.Lemasson B, Serduc R, Maisin C, et al. Monitoring blood-brain barrier status in a rat model of glioma receiving therapy: dual injection of low-molecular-weight and macromolecular MR contrast media. Radiology. 2010;257:342–352. [DOI] [PubMed] [Google Scholar]
  • 19.Pauliah M, Saxena V, Haris M, et al. Improved T(1)-weighted dynamic contrast-enhanced MRI to probe microvascularity and heterogeneity of human glioma. Magn Reson Imaging. 2007;25:1292–1299. [DOI] [PubMed] [Google Scholar]
  • 20.Mills SJ, Soh C, O’Connor JP, et al. Enhancing fraction in glioma and its relationship to the tumoral vascular microenvironment: a dynamic contrast-enhanced MR imaging study. Am J Neuroradiol. 2010;31:726–731. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Yoneda T, Williams PJ, Hiraga T, et al. A bone-seeking clone exhibits different biological properties from the MDA-MB-231 parental human breast cancer cells and a brain-seeking clone in vivo and in vitro. J Bone Miner Res. 2001;16:1486–1495. [DOI] [PubMed] [Google Scholar]
  • 22.Jenkins DE, Hornig YS, Oei Y, et al. Bioluminescent human breast cancer cell lines that permit rapid and sensitive in vivo detection of mammary tumors and multiple metastases in immune deficient mice. Breast Cancer Res. 2005;7:R444–R454. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Heyn C, Ronald JA, Ramadan SS, et al. In vivo MRI of cancer cell fate at the single-cell level in a mouse model of breast cancer metastasis to the brain. Magn Reson Med. 2006;56:1001–1010. [DOI] [PubMed] [Google Scholar]
  • 24.El-Mabhouh AA, Nation PN, Kaddoura A, et al. Unexpected preferential brain metastases with a human breast tumor cell line MDA-MB-231 in BALB/c nude mice. Vet Pathol. 2008;45:941–944. [DOI] [PubMed] [Google Scholar]
  • 25.Song HT, Jordan EK, Lewis BK, et al. Rat model of metastatic breast cancer monitored by MRI at 3 tesla and bioluminescence imaging with histological correlation. J Transl Med. 2009;7:88. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Scheffler K, Lehnhardt S. Principles and applications of balanced SSFP techniques. Eur Radiol. 2003;13:2409–2418. [DOI] [PubMed] [Google Scholar]
  • 27.Bernas LM, Foster PJ, Rutt BK. Imaging iron-loaded mouse glioma tumors with bSSFP at 3 T. Magn Reson Med. 2010;64:23–31. [DOI] [PubMed] [Google Scholar]
  • 28.Miraux S, Massot P, Ribot EJ, et al. 3D TrueFISP imaging of mouse brain at 4.7T and 9.4T. J Magn Reson Imaging. 2008;28:497–503. [DOI] [PubMed] [Google Scholar]
  • 29.Pastel DA, Mamourian AC, Duhaime AC. Internal structure in pineal cysts on high-resolution magnetic resonance imaging: not a sign of malignancy. J Neurosurg Pediatr. 2009;4:81–84. [DOI] [PubMed] [Google Scholar]
  • 30.Xie T, Zhang XB, Yun H, et al. 3D-FIESTA MR images are useful in the evaluation of the endoscopic expanded endonasal approach for midline skull-base lesions. Acta Neurochir (Wien).2011;153:12–18. [DOI] [PubMed] [Google Scholar]
  • 31.Gonzalez-Lara LE, Xu X, et al. In vivo magnetic resonance imaging of spinal cord injury in the mouse. J Neurotrauma. 2009;26:753–762. [DOI] [PubMed] [Google Scholar]
  • 32.Rodriguez O, Fricke S, Chien C, et al. Contrast-enhanced in vivo imaging of breast and prostate cancer cells by MRI. Cell Cycle. 2006;5:113–119. [DOI] [PubMed] [Google Scholar]
  • 33.Moreno H, Hua F, Brown T, et al. Longitudinal mapping of mouse cerebral blood volume with MRI. NMR Biomed. 2006;19:535–543. [DOI] [PubMed] [Google Scholar]
  • 34.Choi JJ, Feshitan JA, Baseri B, et al. Microbubble-size dependence of focused ultrasound-induced blood-brain barrier opening in mice in vivo. IEEE Trans Biomed Eng. 2010;57:145–154. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Townson JL, Ramadan SS, Simedrea C, et al. Three-dimensional imaging and quantification of both solitary cells and metastases in whole mouse liver by magnetic resonance imaging. Cancer Res. 2009;69:8326–8331. [DOI] [PubMed] [Google Scholar]
  • 36.Gril B, Evans L, Palmieri D, et al. Translational research in brain metastasis is identifying molecular pathways that may lead to the development of new therapeutic strategies. Eur J Cancer. 2010;46:1204–1210. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Greenwood J Mechanisms of blood-brain barrier breakdown. Neuroradiology. 1991;33:95–100. [DOI] [PubMed] [Google Scholar]
  • 38.MacDonald IC, Chambers AF. Breast cancer metastasis progression as revealed by intravital videomicroscopy. Expert Rev Anticancer Ther. 2006; 6:1271–1279. [DOI] [PubMed] [Google Scholar]
  • 39.Nagaraja TN, Knight RA, Ewing JR, et al. Multiparametric magnetic resonance imaging and repeated measurements of blood-brain barrier permeability to contrast agents. Methods Mol Biol. 2011;686:193–212. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Palmieri D, Lockman PR, Thomas FC, et al. Vorinostat inhibits brain metastatic colonization in a model of triple-negative breast cancer and induces DNA double-strand breaks. Clin Cancer Res. 2009;15:6148–6157. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Patel S, Dibiase S, Meisenberg B, et al. Phase I clinical trial assessing temozolomide and tamoxifen with concomitant radiotherapy for treatment of high-grade glioma. Int J Radiat Oncol Biol Phys. 2011. [DOI] [PubMed] [Google Scholar]
  • 42.Liu HL, Hua MY, Chen PY, et al. Blood-brain barrier disruption with focused ultrasound enhances delivery of chemotherapeutic drugs for glioblastoma treatment. Radiology. 2010;255:415–425. [DOI] [PubMed] [Google Scholar]
  • 43.Hu J, Ljubimova JY, Inoue S, et al. Phosphodiesterase type 5 inhibitors increase Herceptin transport and treatment efficacy in mouse metastatic brain tumor models. PLoS One. 2010;5:e10108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Nolte I, Brockmann MA, Gerigk L, et al. TrueFISP imaging of the pineal gland: more cysts and more abnormalities. Clin Neurol Neurosurg. 2010; 112:204–208. [DOI] [PubMed] [Google Scholar]
  • 45.Bauknecht HC, Romano VC, Rogalla P, et al. Intra- and interobserver variability of linear and volumetric measurements of brain metastases using contrast-enhanced magnetic resonance imaging. Invest Radiol. 2010;45:49–56. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplemental data
NIHMS1614113-supplement-Supplemental_data.docx (183.4KB, docx)

RESOURCES