Introduction
Patients with advanced non-small-cell lung cancer (NSCLC) commonly develop brain metastases; this occurs in 30–64% of the patients [
1] and results in very poor prognosis. Whereas the survival of individuals with general metastatic NSCLC is reported to be approximately 12 months, the median survival for patients with brain metastases receiving palliative radiotherapy ranges from 2.4 to 4.8 months [
2].
For patients with NSCLC with brain metastases that are symptomatic, the basic treatment approaches are radiation therapy and surgical resection, under the condition that all extracranial lesions are controlled [
1]. On the other hand, when brain metastases in these patients are asymptomatic, systemic chemotherapies (platinum-based chemotherapy and/or molecular targeted drugs) are considered as treatment options, since they are expected to achieve a similar response intracranially as that seen extracranially [
1]. However, the prognosis for survival of these patients continues to be poor, and further improvement is greatly needed.
Bevacizumab, a humanized anti-vascular endothelial growth factor (VEGF) monoclonal antibody (mAb), is approved, based on a high level of evidence, as first-line therapy for metastatic nonsquamous NSCLC in several countries around the world [
3‐
7]. Clinical trials of bevacizumab in patients with recurrent glioblastoma multiforme (GBM) have reported a high response rate [
8,
9]. Following this research on GBM, the studies BRAIN and EOLE, which investigated the use of bevacizumab in patients with nonsquamous NSCLC, demonstrated encouraging efficacy and an acceptable safety profile for chemotherapy that included bevacizumab as a first-line treatment for asymptomatic and untreated brain metastases [
10,
11]. These results indicate that bevacizumab may be useful in patients with NSCLC with brain metastases. One of the specific obstacles to treating brain tumors may be the blood–brain-barrier (BBB), which is generally thought to limit the access of hydrophilic and/or large agents to the central nervous system, CNS [
12,
13]. If so, the BBB blocks an antitumor drug from getting into the brain and targeting brain metastases [
12‐
14]. Therefore, results of clinical studies need to be fully elucidated, specifically by showing that bevacizumab can penetrate the BBB, and by articulating a detailed mechanism for its antitumor effect on metastatic lesions.
Various models of brain metastasis have been reported, including the spontaneous model, the intracranial cellular engraftment model, and the hematogenous model, which respectively require intravenous, intracardial, or intracarotid injections of cancer cells [
15]. Of these, the hematogenous model, which is improved by intracarotid inoculation, has been reported to produce brain metastases from various cancer cell lines efficiently and specifically [
16,
17]. This model and others were used to investigate the role of VEGF signal in brain metastasis, and results suggest that bevacizumab and other VEGF-signaling blockade agents can prevent brain metastasis [
16,
18]. However, it has not yet been confirmed whether bevacizumab has an antitumor effect on already existing brain metastases.
In the present study, we evaluated the antitumor efficacy of bevacizumab on established brain metastases in a xenograft mouse model of hematogenous brain metastases. Using the hematogenous model, which could approximate the status of the BBB seen in clinical settings, we investigated whether bevacizumab would be able to penetrate into metastatic lesions in the brain and exert an anti-angiogenic effect on the formation of tumor microvessels.
Materials and methods
Reagents
Bevacizumab was obtained from F. Hoffmann-La Roche Ltd. (Basel, Switzerland). Human immunoglobulin G (HuIgG) was purchased from MP Biomedicals (Santa Ana, CA, USA). Both bevacizumab and HuIgG were diluted with normal saline. Na-pyruvate and Geneticin™ Selective Antibiotic (G418 Sulfate) were purchased from Thermo Fisher Scientific, Inc. (Waltham, MA, USA). RPMI-1640 culture medium, d-glucose, and HEPES buffer were obtained from Sigma-Aldrich (St. Louis, MO, USA). Fetal bovine serum was obtained from Bovogen Biologicals Pty. Ltd. (Melbourne, Australia).
Construction of pEBMulti-secNluc plasmid for reporter gene
The
XbaI (blunt-ended)/
KpnI secreted NanoLuc® fragment [
19] was removed from pNL2.3[secNluc/Hygro] (Promega Corporation, Madison, WI, USA) and ligated into the
EcoRV/
KpnI sites of pEBMulti-Neo (Wako Pure Chemical Industries, Ltd., Osaka, Japan), resulting in the generation of the pEBMulti-secNluc plasmid.
Cell lines and culture conditions
The human NSCLC cell line NCI-H1915, which was originally isolated from a brain metastasis [
20], was obtained from American Type Culture Collection (Manassas, VA, USA) and maintained in RPMI-1640 medium supplemented with 10% fetal bovine serum, 4.11 g/l
d-glucose, 10 mM HEPES buffer, and 1 mM Na-pyruvate at 37 °C in humidified air with 5% CO
2. The Nluc-H1915 cell line, which stably expresses the luciferase reporter gene secNluc, was prepared by transfecting pEBMulti-secNluc plasmid into NCI-H1915 cells using FuGENE® HD Transfection Reagent (Promega Corporation), and then stable cells were selected and maintained in the culture medium containing 200 µg/ml of G418 Sulfate. The number of cells was measured using NucleoCounter® NC-200™ (ChemoMetec A/S, Allerod, Denmark).
Laboratory animals
Seven-week-old male C.B-17/Icr-scid/scidJcl mice and CB17/Icr-Prkdc < scid > CrlCrlj mice (SCID mice) were obtained from CLEA Japan, Inc. (Tokyo, Japan) and Charles River Laboratories Japan, Inc. (Kanagawa, Japan), respectively. All animal experiments were reviewed and approved by the Institutional Animal Care and Use Committee at Chugai Pharmaceutical Co., Ltd., and conformed to the Guide for the Care and Use of Laboratory Animals published by the Institute of Laboratory Animal Resources.
Analysis of Nluc activity
Luminescence was measured using Nano-Glo® Luciferase Assay System (Promega Corporation), which provides a simple, single-addition reagent that generates a glow-type signal in the presence of NanoLuc® luciferase. The reagent was prepared by mixing Nano-Glo® Luciferase Assay Substrate with Nano-Glo® Luciferase Assay Buffer according to the manufacturer’s instructions. 2104 EnVision™ Multimode Plate reader (Perkin Elme, Inc., Waltham, MA, USA) was used to measure the luminescence.
Evaluation of antitumor activity in a subcutaneous Nluc-H1915 model
SCID mice were subcutaneously inoculated with 5 × 10
6 cells/mouse of Nluc-H1915 cells into the right flank and, when tumor volume (TV) reached 223–403 mm
3, were randomized into control and test groups that received HuIgG and bevacizumab, respectively (day 1). TV was estimated from the equation TV = ab
2/2, where a and b are tumor length and width, respectively. Bevacizumab or HuIgG was intraperitoneally administered once a week at a dose of 5 mg/kg (days 1, 8, and 15), based on the regimen in a previous report [
21]. To evaluate the antitumor activity, TV and Nluc activity were measured on day 22. Nluc activity [relative light unit (RLU)/whole tumor] was measured in the supernatant of the tumor lysate, which was prepared by homogenizing the tumor specimen in a cell lysis buffer (Cell Signaling Technology, Inc., Danvers, MA, USA).
To produce brain metastases, SCID mice were anesthetized by isoflurane and a microclamp was applied to the external carotid artery to prevent extracranial metastasis from forming. Then Nluc-H1915 cells (1 × 105/50 µl RPMI-1640) were injected slowly into the internal carotid artery by surgical visualization, and finally the cut made in the skin was stitched up. Tumor inoculation was typically performed on 40 mice and approximately 10–30% of the mice died or were euthanized due to loss of locomotor activity before randomization. Approximately 10% of the mice didn’t develop brain metastasis based on an assessment of plasma Nluc activity and the rest of the mice were used for experiments.
After 15–17 days, the Nluc activity in plasma (RLU/5 µl) was measured and the plasma levels
were used to randomize mice into control and test groups to obtain similar levels of average plasma Nluc activity in each group (day 1). Then, 5 mg/kg of HuIgG and bevacizumab were administered intraperitoneally on days 1, 8, 15 into mice in the control and test groups, respectively. On day 21, the antitumor activity was evaluated by measuring the Nluc activity (RLU/whole brain) in the supernatant of the lysate of brain parenchyma homogenized with cell lysis buffer. The presence of brain metastases was examined under a microscope after hematoxylin and eosin (HE) staining or immunohistochemistry (IHC) with an anti-human EGFR antibody. The number of metastatic lesions having a maximum diameter of 300 µm or more was a minimum of 3 in our IHC analysis in control HuIgG-treated group.
Tumor tissues or brains were removed from mice on the specified day, frozen immediately in liquid nitrogen, and stored at − 80 °C until use. The tumor tissues or the brain parenchyma cells were homogenized in cell lysis buffer. The supernatant in each homogenate was collected after centrifugation and used for analyses.
Measuring human VEGF and human IgG by enzyme-linked immunosorbent assay (ELISA)
After perfusion with saline, brains were removed and the concentrations of human VEGF and bevacizumab in the tissue samples of mice were determined according to the manufacturer’s instruction using a Human VEGF ELISA Kit (ab100663) and a HuIgG ELISA Kit (ab100547) (Abcam plc, Cambridge, UK), respectively. Benchmark Plus Microplate Reader (Bio-Rad Laboratories, Inc., Hercules, CA, USA) was used to detect the concentration of HuIgG in the tissue sample as an indication of the level of bevacizumab.
Determining human EGFR and human IgG in tumor tissues by IHC
The mice were euthanized at the specified time points by exsanguination under anesthesia using isoflurane. Brains were removed without perfusion and sliced into six sections. Human NSCLC cells in the brain were stained by IHC with an anti-human EGFR antibody, EGFR (D38B1) XP® Rabbit mAb (Cell Signaling Technology, Inc.). Bevacizumab was stained by IHC using Polink-2 Plus HRP HuIgG with DAB Kit (Golden Bridge International, Inc., Bothell, WA, USA).
Quantification of tumor blood vessels in brain metastases
Microvessel density (MVD) in the metastatic lesions in brain was assessed by IHC with rat anti-mouse CD31 mAb (BD Biosciences, San Jose, CA, USA). MVD (%) was calculated from the ratio of the CD31-positive staining area to the total observed brain metastasis area in the viable lesion. Positive staining areas were calculated using the imaging analysis software (Definiens Tissue Studio®; Definiens AG, Munich, Germany).
Statistical analysis
Differences between two groups were assessed using Student’s t-test, or Wilcoxon’s rank sum test. All reported p values were two-sided, and those < 0.05 were considered to be statistically significant. Data are represented as mean and SD. All statistical analyses were conducted using the JMP® Version 11 software (SAS Institute, Inc., Cary, NC, USA).
Discussion
In this study, we demonstrated the anti-proliferative effect of bevacizumab on established brain metastases in a hematogenous brain metastasis xenograft model using the human NSCLC cell line, Nluc-H1915. We used NCI-H1915 cells to develop this xenograft model because they were originally isolated from a brain metastasis in a patient with NSCLC, and tumor spheres of these cells have been reported to form multifocal brain tumors when intracranially injected into NOD-SCID mice [
5,
22]. Of note, we chose to inject the tumor cells through the internal carotid artery in these mice, because the intracarotid artery injection of tumor cells recreates the latter part of hematogenous brain metastasis, e.g. adhesion to the brain vasculature, extravasation, and outgrowth in a brain-specific environment that supplies oxygen and nutrients [
9], while clamping the external carotid artery is expected to further reduce metastases to other organs. Because these characteristics reflect the clinical setting for NSCLC patients with brain metastases, this model is suitable for practically evaluating the antitumor effect of bevacizumab on established brain metastases. In addition, this method can also be used to investigate earlier steps—when metastatic nodules form in the vascular niche—and for evaluating the drug’s effect on adhesion to the brain vasculature and extravasation.
It is difficult to quantify brain metastases, especially when they have a multifocal profile. The Nluc-H1915 cell line established in this study stably expresses the Nluc reporter gene and can also be transplanted into SCID mice. Nluc activity in the lysates of tumors grown in these mice correlated significantly with TV at a high R2 value, and the antitumor activity of bevacizumab in mice subcutaneously transfected with Nluc-H1915 could be assessed. Thus, Nluc-H1915 cells allowed us to estimate brain metastasis tumors semi-quantitatively, using a lysate of the whole brain parenchyma including metastases.
Regimens that contain bevacizumab are recommended as standard first-line therapy for metastatic NSCLC. Bevacizumab has recently been reported to have encouraging efficacy in patients with NSCLC who have asymptomatic untreated brain metastases, as shown by the phase 2 BRAIN study, with activity shown in both extracranial and intracranial lesions [
1,
10,
11,
23]; however, its penetration into brain lesions and the mode of action have not yet been fully investigated. Specific features of the BBB have been reported to affect systemic treatment of metastatic tumors in the brain [
24,
25], since hydrophilicity and the large molecular size of an antibody are considered disadvantageous for penetration. Tumor cells injected through the internal carotid artery in our mouse model reproducibly formed metastatic nodules in the brain, which theoretically suggests that the character of the BBB in this model is similar to those of patients with brain metastasis. The mean concentration of bevacizumab in the brain parenchyma was higher in the brain metastasis model mice than in normal mice (Fig.
4a), and when assessed by IHC, bevacizumab localized inside blood vessels in the normal brain parenchyma of metastatic mice as well as in normal mice. However, it did specifically penetrate into the brain metastatic lesions (EGFR+) caused by hematogeneous metastases, which is absolutely required for this antibody to inhibit the extravascular VEGF produced by tumor cells. As a result of this intratumoral penetration, bevacizumab worked well in reducing MVD. This suggests that bevacizumab could penetrate into tumor tissues because the BBB in the metastatic brain lesions is altered, whereas in normal mice it is intact. Indeed, a study by Oosting et al. of PET scans of patients with metastatic renal cell carcinoma given
89Zr-bevacizumab suggested that bevacizumab accumulates in metastatic brain lesions but not in normal brain parenchyma [
26], though it was uncertain whether there was enough accumulated bevacizumab to provide efficacy. Taking our result and this report together, it is highly likely that the clinical efficacy of bevacizumab in patients with NSCLC with brain metastases reported in the BRAIN and the EOLE studies [
10,
11] was achieved because bevacizumab was able to penetrate the altered BBB and enter metastatic lesions.
Recent studies have reported the efficacy of other therapeutic antibodies that inhibit immune checkpoints in patients with asymptomatic brain metastases [
27,
28], which provides further evidence for the penetration of IgG into metastatic brain tissue and supports our findings. The precise status of BBB in metastatic brain nodules and the mechanism of penetration and/or transportation of IgG across the BBB should be investigated further, both clinically and in preclinical studies.
Bevacizumab inhibits VEGF-induced neovascularization, which is required for tumor growth, and partially normalizes the abnormal hyper-permeability caused in tumor vessels by VEGF-A. This normalizing effect decreases interstitial fluid pressure in the tumor and improves drug delivery into tumor tissues [
29]. In this study, Nluc-H1915-derived VEGF was detected in the brain, bevacizumab was found distributed in brain metastatic lesions (Fig.
4d–f), and MVD in the brain metastasis was significantly lower in mice treated with bevacizumab than in HuIgG-treated mice (Fig.
5c). The inhibition of neovascularization appears to be at least one of the major modes of action for the anti-proliferative effect of bevacizumab on brain metastases in this model.
In conclusion, we demonstrated the anti-proliferative effect of bevacizumab on established brain metastasis in the hematogenous brain metastasis xenograft model. The distribution of bevacizumab in metastatic lesions suggests that it penetrated the BBB, and the reduction in MVD suggests that it suppressed angiogenesis, which indicates that the anti-proliferative effect of bevacizumab on brain metastasis observed in this study is due to its anti-angiogenic activity. The mechanism by which HuIgG penetrates the BBB in the hematogenous metastatic model and/or in patients with brain metastasis needs to be elucidated in more detail. The outcomes of this study suggest that bevacizumab has efficacy against established brain metastases, and that a regimen containing bevacizumab could be a promising treatment option for patients with NSCLC with brain metastases.
Compliance with ethical standards
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.