Background
Glioblastoma (GBM) tumors are largely refractory to systemic treatments; the median survival time for patients with GBM is 10 months and the 2-year survival rate is less than 10%. Chemotherapy for GBM is compromised in part by the blood-brain barrier limiting drug access to the malignant cells. In addition, pre-clinical models showed that GBM tumors are poorly perfused [
1,
2] due to factors such as reduced blood flow rates, elevated hematocrit and interstitial fluid pressure, and an increase in geometric resistance [
3‐
6], all of which impede drug delivery to the tumor tissue. Strategies which improve vascular function in GBM tumors should improve the delivery of other drugs capable of crossing the blood brain barrier and this should be associated with an increase in therapeutic activity.
Our laboratory has previously characterized and described the effects of a liposomal formulation of irinotecan (Irinophore C™) [
7,
8]. Encapsulation of irinotecan into liposomes improved the pharmacokinetic profile of the drug and its active metabolite, SN-38. More specifically, administration of Irinophore C™ resulted in a 1000-fold increase in the area-under-the-curve of plasma irinotecan concentration when compared to free drug (Camptosar). In addition, following irinophore C™ injection, the plasma levels of SN-38 were maintained at concentrations that were up to 40-fold higher than that achieved following injection of free drug [
7]. Following irinophore C™ treatment, the s.c. (subcutaneous) colorectal tumors (HT-29) exhibited more functional tumor blood vessels, reduced hypoxia, and increased tumor perfusion. Importantly, these changes in tumor vasculature were associated with increased tumor uptake of doxorubicin and 5-FU given intravenously [
8]. The latter data were consistent with the idea that the tumor vasculature in the treated tumors acquires a more "normal-like" function; an effect of anti-angiogenic therapies described as 'normalization' [
9,
10].
The primary goal of the studies reported here was to determine whether Irinophore C™ is efficacious in models of GBM, and whether treatment with this drug formulation would also result in normalization of GBM vasculature. The effects of Irinophore C™ on the growth rates and vascular function of the HT-29 colorectal cancer model was attributed to significant increases in the drug circulation lifetime and plasma concentration when encapsulated in liposomes [
7,
8]. We further reasoned that liposomal formulations of other drugs with known activity against proliferating endothelial cells should have preferential cytotoxicity towards angiogenic tumor vessels and could potentially also 'normalize' the chaotic and erratic vasculature of tumors. Thus, part of these studies assessed the effects of liposomal vincristine [
11] and doxorubicin (Caelyx
®) on tumor vasculature. Vincristine has previously been shown to be active against proliferating endothelial cells [
12]. Liposomal formulations of doxorubicin have also been shown to have direct effects on tumor associated vasculature [
13‐
15].
The data reported here assess the effects of Irinophore C™, Caelyx® (a commercially available and FDA-approved liposomal formulation of doxorubicin), and liposomal vincristine on tumor vasculature in subcutaneous and orthotopic models of GBM. The results indicate that Irinophore C™ was the most active formulation when using treatment endpoints based on changes in tumor size as well as tumor vascular morphology and function in GBM grown subcutaneously and orthotopically. The effects were consistent with the idea that following treatment, there was normalization of tumor vasculature. In the subcutaneous tumors, vascular 'normalization' was associated with increased tumor uptake of Hoechst 33342, while in the orthotopic glioma tumors, treatment-induced vascular 'normalization' was associated with decreased tumor uptake of Hoechst 33342.
Methods
Cell culture
Adult dermal human microvascular endothelial cells (d-HMVEC; Cambrex Bio Science, Walkersville, MD), Human brain microvascular endothelial cells (HBMEC; ScienCell Research Laboratories, San Diego, California) and U251MG glioblastoma cells (American Type Culture Collection, Manassas, VA) were characterized and authenticated by the cell banks using immunofluorescent methods and used for a maximum of eight passages for the endothelial cells and fifteen passages for U251MG. Stock cells lines were maintained in the absence of penicillin and streptomycin and screened for mycoplasma prior to preparing a stock of cells that were frozen for use in experiments. D-HMVEC cells were maintained in Endothelial Cell Basal Medium-2 (Clonetics®, Lonza, Basel, Switzerland) supplemented with 5 ng/mL Fibroblast Growth Factor, 20 ng/mL Vascular Endothelial Growth Factor, 10 ng/mL Epidermal Growth Factor (Clonetics®, Lonza), 10 unit/mL Heparin (Pharmaceutical Partners of Canada) 1% L-glutamine, 1% penicillin/streptomycin (Stem Cell Technologies, Vancouver, BC, Canada) and 10% Fetal Bovine Serum (FBS; Hyclone, Logan, UT), and plated in 1% gelatin (Sigma, Oakville, ON, Canada) pre-coated dish. HBMEC cells were maintained in Endothelial Cell Medium supplemented with Endothelial Cell Growth Supplement (ScienCell Research Laboratories) containing 5 μg/mL Insulin, 10 ng/mL Epidermal Growth Factor, 2 ng/mL Fibroblast Growth Factor, 2 ng/mL Insulin-like Growth Factor-1, 2 ng/mL Vascular Endothelial Growth Factor, 1 μg/mL hydrocortisone, 5% FBS and 1% penicillin/streptomycin, and plated in 15 μg/mL fibronectin (Sigma) pre-coated dish. U251MG cells were maintained in DMEM medium supplemented with 1% L-glutamine, 1% penicillin/streptomycin (Stem Cell Technologies, Vancouver, BC, Canada) and 10% FBS (Hyclone, Logan, UT). All cell lines were cultured at 37°C in a humidified atmosphere containing 5% CO2, and used during exponential growth phase unless otherwise stated.
GBM animal model s.c. and orthotopic
All protocols involving work with live animals were reviewed and approved by the University of British Columbia Animal Care Committee (certificate of approval # A07-0423). For the subcutaneous GBM model, U251MG cells (5 × 10
6) were implanted subcutaneously into the backs of Rag2M mice (7-10 weeks old females, n = 9). To generate orthotopic GBM tumors, U251MG (7.5 × 10
4) cells were implanted into the right caudate nucleus-putamen (ML -1.5 mm; AP +1 mm; DV -3.5 mm) of mice (n = 5-6) using a stereotaxic injection frame (Stoelting Company, Wood Dale, IL). Animals were treated with 25 mg/kg Irinophore C™, 2 mg/kg liposomal vincristine or 15 mg/kg doxorubicin liposome (Caelyx
®, Schering-Plough, QC, Canada) i.v. on day 21, 28 and 35 after inoculation. Dosing of liposomal vincristine and Caelyx
® resulted in less than 5% body weight loss, while Irinophore C™ treatment did not cause any change in body weight. Previous tests in our laboratory have shown that the maximum tolerated single doses for Irinophore C™, Caelyx
® and liposomal vincristine are >120 mg/kg, 17 mg/kg and 3 mg/kg, respectively. Irinophore C™ [
16] and liposomal vincristine [
17] were prepared as described previously. S.c. tumor size was measured throughout the study by caliper and tumor weights were extrapolated from the measurements using the following formula: mg = (tumor width^2 × tumor length)/2 [
18]. Mice were injected with Hoechst 33342 (1.2 mg/mouse; Sigma) twelve (s.c. model) or twenty (orthotopic model) minutes prior to sacrifice on day 42. This timing was chosen based on previous study [
8] and tests (not shown) aimed at determining the optimal timing for Hoechst 33342 injection without saturation of the tissue and before any decrease in Hoechst 33342 staining could be observed due to possible metabolic elimination. All animals were terminated by CO
2 asphyxiation and s.c. tumors or brains were harvested and cryopreserved in OCT (Sakura Finetek, CA) on dry ice and stored at -80°C.
Hoechst 33342, Ki67, CD31, VEGFR2, EF5, Collagen IV, NG2 and nuclei density staining and quantification
Optimal Cutting Temperature compound (OCT)-preserved s.c. tumors were cryosectioned using a Leica CM1850 Cryostat (Leica, ON, Canada) and 10 μm sections were collected in the middle of each tumor. OCT preserved brains were cryosectioned and 10 μm sections were collected from the Bregma +1.0 location. Sections were fixed in a 1:1 mixture of acetone:methanol for 15 minutes at room temperature, then blocked with blocking buffer (Odyssey blocking buffer, Rockland, PA) for 1 hour at room temperature. Sections were stained with rat anti-mouse CD31 antibody (1:100 dilution, PharMingen #550274, BD Biosciences), rabbit anti-human Ki-67 (Invitrogen #18-0191z; 1:100), rabbit anti-human/mouse vascular endothelial growth factor receptor 2 antibody (VEGFR2; 1:100; Cell Signaling technology #2479, NEB, Pickering, ON, Canada), rabbit anti-Collagen IV antibody (1:400, Abcam # ab19808, Cambridge, MA) and mouse anti-NG2 chondroitin sulfate proteoglycan antibody (1:100, Millipore # MAB5384, Billerica, MA). Primary antibodies were incubated on sections overnight at 4°C. Secondary antibodies (Alexa 488 goat anti-rat #A11006, Alexa 546 goat anti-rabbit #A-11035 and Alexa 633 goat anti-mouse #A-21126, 1:200, Invitrogen) were incubated for 1 hr at room temperature. Nuclei were stained with Draq5 (Biostatus, Leicestershire, UK; 1:200) for 30 min at 37°C. Slides were mounted with PBS and imaged for Alexa 488 (L5 filter), Hoechst 33342 (A4 filter), Alexa 546 (Cy3 filter), Cy5 (Cy5 filter) and Draq5 (Cy5 filter) using a robotic fluorescence microscope (Leica DM6000B, Leica, ON, Canada) and a composite color image of these markers was produced (Surveyor software, Objective Imaging Ltd.). Thresholds for each marker were set using Photoshop; the threshold level was set using a scale from 1 to 255 units, and was defined at 2 units higher than the minimal level necessary to obtain a negative signal for non-specific staining, and was kept the same for all sections. Acquired images were quantified for positive pixels or colocalization (double-positive pixels) using an in-house segmentation algorithm, normalized to the number of pixels in the tumor area and expressed as positive fraction (positive pixels divided by non-necrotic tumor area; MATLAB, The Mathworks, Natick, MA). Non-necrotic tumor areas were defined by cropping out necrotic and non-tumor tissue on the basis of positive Ki-67 and Draq5 co-stained sections and were quantified using the same in-house algorithm. Colocalization was considered positive when two positive pixels from one stain of interest were located within a 3 pixels radius from one pixel of the other stain of interest. Of note, one cell nucleus measures between 3 and 6 pixels. Blood vessel diameter was defined by taking 10 measurements/tumor section in a 15 × 15 cm box at 200% magnification using Photoshop, and was expressed in pixels. For differential analysis between the tumor's center and periphery, the boundary between the tumor center and periphery area was established at 20% of tumor diameter distance from tumor margin. Another set of sections was stained with hematoxylin and eosin for histopathology analysis. The fraction of collagen IV-free blood vessels was defined as Collagen IV negative/CD31 positive pixels over total CD31 pixels. The fraction of NG2-free blood vessels was defined as NG2 negative/CD31 positive pixels over total CD31 pixels. The amount of basement membrane empty sleeves was defined as CD31 negative pixels/collagen IV positive pixels divided by the total non-necrotic tumor area.
Magnetic Resonance Imaging and Ktransmeasurement in U251MG orthotopic tumors
All magnetic resonance experiments were carried out using a 7.0 Tesla MR scanner (Bruker, Ettlingen Germany). A Bruker (Ettlingen, Germany) volume coil (inner diameter of 7 cm) and rectangular surface coil (1.7 × 1.4 cm) was used for signal transmission and reception respectively. The coil was tuned to the hydrogen proton frequency (300.3 MHz). The K
trans values were obtained from serial images acquired to monitor changes in the concentration of a MR-visible contrast agent (GD-DTPA; Bayer Schering Pharma) within each pixel, during the initial uptake and subsequent washout of the agent in the tumor. The MRI scans follow the protocol reported by Lyng et al. [
19]; briefly, mice were anaesthetized with isofluorane (5% induction, 2% maintenance), a catheter inserted into the lateral tail vein and the animal was placed supine with its head above the surface coil. A proton-density weighted scan was first acquired to serve as a baseline for conversion of pixel intensity to absolute concentration values of the contrast agent. A volume equivalent to 10 uL per gram body weight of the contrast agent (0.03 M Gd-DTPA in saline) was injected via the tail vein catheter in a period of 10-15 seconds. The contrast series consisted of a 3D RF-spoiled Fast Low Angle Shot (FLASH) sequence with timing and resolution parameters as follows: echo time/repetition time = 2.8/9.2 ms, Field of view = 1.92 × 1.92 × 1.6 cm, Matrix size = 128 × 128 × 16 cm, acquisition time per image = 9.45 seconds. Twenty baseline scans were acquired before contrast agent injection and 250 scans were acquired afterwards, resulting in a total acquisition time of 43 minutes. The concentration-time curve for each pixel was fit to a two-compartment Kety model [
20] which describes the pharmacokinetics of the contrast agent using three parameters: ve (volume of extracellular extravascular space), K
trans (volume transfer constant between the vasculature and tissue compartment) and Vp (fractional volume of the vascular compartment).
In vitroendothelial cell exposure and nuclei count
For proliferative conditions, Dermal Human MicroVascular Endothelial Cells (d-HMVEC; 600 cells/well) and Human Brain Microvascular Endothelial Cells (HBMEC; 5000 cells/well) were plated in black 96-well plates (Optilux™, BD Biosciences, Mississauga, ON, Canada) and drugs were added the day after. For non-proliferative conditions, d-HMVEC cells (5000 cells/well) and HBMEC (50000 cells/well) were plated in black 96 well plates and drugs were added four days after. Irinotecan (Sandoz, QC, Canada), SN-38 (LKT Laboratories, MN, USA), vincristine (Novopharm, ON, Canada), docetaxel, paclitaxel (Taxol®; Bristol Myers Squibb Canada, QC, Canada) and doxorubicin (AdriamycinTM/MC, Pfizer, QC, Canada) were added in concentrations ranging from 1-100,000 picoMolar on cells and replaced daily for 7 days. At the end of drug treatment, cells were fixed with 3.5% paraformaldehyde (Electron Microscopy Sciences, PA) for 15 minutes at -20°C, permeabilized with 0.1% Triton (Perkin-Elmer, MA) in PBS for 10 minutes at room temperature, blocked for 1 hr at 4°C (Odyssey blocking buffer, Rockland, PA) and incubated overnight with Ki67 antibody (Invitrogen #18-0191z; 1:100 dilution in blocking buffer). Cells were then incubated with Anti-rabbit Alexa 488 secondary antibody (Molecular Probe #A11034, Invitrogen; 1:200 in blocking buffer) for 1 hr at room temperature. Nuclei were stained with Draq5 dye (Biostatus, Leicestershire, UK; 1:200 in PBS) for 30 min at 37°C. Twenty fluorescent photographs/well (Alexa 488 emission: 475 nm, excitation: 535 nm; Draq5 emission: 620 nm, excitation: 700 nm) were taken at 10 × magnification using an InCell Analyzer 1000 (Amersham Bioscience) and the total nuclei count (Draq5 stained nuclei) as well as Ki67 expressing nuclei count (Draq5 and Alexa 488 double stained nuclei) were quantified using InCell Developer Toolbox software (Amersham Bioscience, GE Healthcare, Baie d'Urfe, QC, Canada). Dose-response curves generated from total nuclei count were used to calculate drug concentrations causing a decrease in endothelial cell nuclei count by 20% (fraction affected: Fa = 0.2), 50% (Fa = 0.5), 75% (Fa = 0.75) and 90% (Fa = 0.9) and compared for both proliferative and non-proliferative cells. All data points represent the average of 3 independent experiments in triplicate +/- S.E.M.
Statistical analysis
All statistical data was collected using GraphPad Prism (San Diego, CA). Because all treatment drugs were chosen based on previous rationale justifying their inclusion in the study, the experimental design should not be regarded as a screening assay and statistical analysis was done using the single comparison non-parametric two-tailed Mann Whitney test and no correction was made for multiple comparisons. All data are expressed +/- S.E.M.
Discussion
Studies over the last few decades have established that liposomal formulations of selected antineoplastic agents can be more effective than the same drug administered in free form. Liposomal formulations of anticancer drugs are known to have long circulation half-lives
in vivo, and release the drug slowly over time [
7,
11]. Thus, the pharmacological properties of a drug given in its free form (e.g. via bolus injection or slow infusion) is changed dramatically by encapsulation in liposome. As a result, one might anticipate that the use of liposomal drugs will expose tumors to drugs for extended periods of time when compared to treatment with the free drug. This, of course, is well established in the literature and has been explained on the basis of the enhanced permeability and retention effect known to promote accumulation of intravenously administered liposomal drug formulations in tumors [
34]. What is often not considered in studies with liposomal formulations is that these formulations constantly release the associated drug while in the circulation compartment, thereby extending the presence of the drug in the plasma compartment. This study tries to address whether part of the treatment benefits could be attributed to direct effects of the free drug (available in the blood compartment) on tumor vascular endothelial cells. The fact that these drug formulations are active against proliferating vasculature was anticipated, but not demonstrated to date. Liposomal drug formulations are known to accumulate and release drugs in close proximity to tumor blood vessels [
14,
15]. More intriguing, however, is the possibility that exposing the tumor vasculature to low concentrations of drug for extended periods may produce effects that are comparable to the vascular normalization effects described in the context of anti-angiogenic therapy [
9,
10] as discussed below.
In the present study, it is demonstrated that Irinophore C™, Caelyx
® and liposomal vincristine are effective against GBM grown subcutaneously or orthotopically (in the brain). The tumor masses in treated animals were significantly smaller compared to control (p < 0.001; Figure
1a), indicating that the liposomal drugs used in this study are potent against GBM, regardless of the site of tumor growth. Analysis of the tumor tissue, and in particular the vascular morphology, also indicates that treatments affected the tumor vasculature to various degrees. Overall, Irinophore C™ impacted the vasculature to a greater extent than the other formulations, and generated tumors with blood vessels that were morphologically more mature. In the subcutaneous model, Irinophore C™ restored the basement membrane architecture, increased the pericyte coverage and reduced blood vessel diameters. The data suggest a restoration of the vessel architecture to a more normal state. In the more clinically relevant orthotopic model, Irinophore C™ treatment restored the basement membrane architecture and reduced blood vessel diameters of the tumor vasculature, again suggesting a restoration of the vessel architecture to a more normal state. Irinophore C™ treatment also increased the quantity of vessel staining in the center of tumors, suggesting a more homogenous distribution of blood across the entire tumor. Further, Irinophore C™ reduced K
trans values calculated from Dynamic Contrast Enhanced (DCE)-MRI studies significantly. Based on changes in vessel morphological appearance, the drop in K
trans values was interpreted as a decrease in vessel permeability [
35], and is consistent with the suggestion that Irinophore C™ treatment improved vascular function in the tumor. The larger variability in K
trans values determined in tumors from control animals reflects the random nature of chaotic and leaky blood vessels in individual tumors [
36]. It had already been established in s.c tumors that Hoechst 33342 could be used as a marker for tumor vessel function by validation with K
trans measurements [
8], but this had not been done for the orthotopic GBM tumor described here. It is shown here that the observed reduction in Hoechst 33342 staining after treatment while total CD31 staining remained constant correlates with a reduction in K
trans measures. Taken together, these observations strongly suggest an improvement in vascular function. The tumor blood vessels in tumors from animals treated with Irinophore C™ behave more like vessels in the normal brain where the blood-brain barrier is intact.
The concept of 'blood vessel normalization' was first postulated in the 70s [
37] and more recently, the clinical potential of vascular normalization has been described [
9,
10]. As with most solid tumors, the microvasculature of gliomas is characterized by tortuous and fenestrated vessels with diameters that are larger than normal [
38] and discontinuous basement membrane which rarely encloses pericytes [
39]. In glioma [
28,
29,
40], antiangiogenic therapies can stop the growth of tumor vessels, prune immature and inefficient tumor vessels and normalize surviving vasculature by increasing the fraction of pericyte-covered vessels, restoring the abnormally thick and irregular basement membrane and reducing the high vascular permeability of these vessels [
9,
10]. In glioblastoma patients, a "vascular normalization index" was defined by changes in vascular permeability (K
trans values), microvessel volume and circulating collagen IV. It was found that this index was closely associated with overall survival and progression-free survival in response to Cediranib, a pan-VEGFR inhibitor [
40]. Pre-clinically, the delivery of temozolomide in an intracerebral model of glioma increased after treatment with the angiogenesis inhibitor SU5416. This drug restored capillary architecture and decrease interstitial fluid pressure [
41]. Such studies offer strong evidence that the tumor vasculature in GBM is a valid target, and that therapies which 'normalize' tumor vasculature may improve the delivery of a second drug at some point in the treatment regimen.
The studies described here, together with an earlier publication [
8], offer strong evidence that liposomal formulations of selected drugs, and especially Irinophore C™, induce a normalization of the tumor vasculature. In this study, collagen IV and NG2 were used as markers for basement membrane and pericytes, respectively. However, there is no consensus in the field for a definitive marker of these parameters. Other markers used to evaluate basement membranes include nidogen or laminin, and desmin or α-smooth muscle actin for pericytes [
9,
30]. These caveats notwithstanding, the morphological changes observed were associated with changes in Hoechst 33342 uptake in the tumor and when using this parameter, remarkably different results were obtained depending on the site of tumor growth (subcutaneous vs orthotopic). In the subcutaneous model, the liposomal treatments increased the amount of Hoechst 33342 staining in the tumor tissue (Figure
1c), while in the orthotopic tumors Hoechst 33342 staining was reduced (Figure
2c). As noted above, treatment effects were similar if blood vessel morphology parameters were used as a measured endpoint. While initially surprising, the Hoechst 33342 uptake data may actually be consistent with restoration of the blood-brain barrier, which is more impermeable to Hoechst 33342. It is well established that Hoechst 33342 is a p-glycoprotein substrate [
42]. It does not accumulate in normal brain tissue because it cannot cross the blood brain barrier, but it is present in untreated orthotopic brain tumors which exhibit leakier blood vessel. This idea was further confirmed by K
trans measurements, which strongly suggested a vasculature normalization induced by Irinophore C™. This interpretation suggests that Hoechst 33342 is not an appropriate marker for tumor
perfusion in orthotopic glioma models, as it was previously used in a s.c. tumor model [
8]. It does, however, function as a
permeability marker for perfused tumor associated blood vessels, which is reduced upon normalization. The impact of vascular normalization on tumor perfusion in orthotopic GBM tumors could not be assessed in the present study because MRI K
trans data and Hoechst 33342 staining data are not direct measures of perfusion in the brain tumor. However, data obtained in the subcutaneous model suggest that treatment with liposomal drugs does not reduce tumor perfusion, as measured by CD31/Hoechst 33342 double staining, and may even increase it, as suggested by data obtained from Caelyx
®-treated s.c. tumors. Studies to measure the delivery of a second drug that can cross the BBB in liposomal drug-treated tumors are underway and will provide an indication of the impact of vascular normalization on vessel perfusion in the orthotopic model.
The idea that liposomal formulations of anti-cancer drugs, in addition to having a direct cytotoxic effect on the tumor cells, may also act as through anti-angiogenic mechanisms is intriguing. It seems reasonable to suggest that the extended drug release characteristics associated with the liposomal drug formulations used in this study [
7,
11] may have effects on blood vessels in a manner similar to metronomic dosing schedules - i.e. frequent, low dose administration of drugs with no prolonged drug-free breaks [
43]. Metronomic dosing is now acknowledged to act specifically on the proliferating endothelial cells of tumor blood vessels [
44] and was more recently shown to improve tumor perfusion and to decrease hypoxia in a pancreatic tumor model [
45]. To examine this hypothesis, an
in vitro assay was used to evaluate the activity of irinotecan, doxorubicin and vincristine (the drugs encapsulated by liposome examined in this study) against proliferating endothelial cells. The assay was adapted from one developed by Bocci et al. to examine the effects of metronomic drug exposure against endothelial cells [
33]. Previous reports suggest that docetaxel and paclitaxel have potent activity against endothelial cells in an
in vitro metronomic dosing regime [
32,
33,
46], so these drugs were included in the assay as positive controls. The effects of SN-38 were also evaluated in the assay because SN-38 is a more active metabolite of irinotecan generated by tissue and plasma carboxylesterases
in vivo [
47,
48]. Further it has already been established that following treatment with Irinophore C™, high levels of SN-38 are maintained in the plasma compartment for extended time periods [
7]. SN-38 levels may play an important role in the anti-cancer activity of Irinophore C™.
The
in vitro metronomic dosing assay presented in Figures
7 and
8 suggest that vincristine and SN-38, like the taxanes (docetaxel or paclitaxel), are highly active against proliferating endothelial cells (Figure
6a-b). In contrast, free irinotecan has little specificity for proliferating endothelial cells over non-proliferating cells
in vitro. The data for free vincristine corroborate the effects on tumor vasculature seen with the liposomal form of the drug used here, while the results obtained with free irinotecan, which is not specific for proliferating endothelial cells, is actually contradictory. Irinophore C™ was the most active of the three liposomal formulations used. The results in Figure
7 and
8 would strongly suggest that the activity of Irinophore C™may be explained by the high plasma levels of SN-38 generated following administration of the formulation [
7,
16]. Thus it can be concluded from the studies presented here that the active metabolite of irinotecan, SN-38, may be the agent promoting vascular normalization in the models used here.
Interestingly, the
in vitro assay suggests that doxorubicin should have little specificity on proliferating endothelial cells, yet i.v. administration of Caelyx
® resulted in effects on the tumor vasculature that were comparable to those seen following administration of Irinophore C™. The reasons for this are unclear at present but may be related to disruptions in the production of hypoxia-induced VEGF caused by doxorubicin [
49]. Previous studies completed using the rat intracranial 9L tumor model treated with a formulation of doxorubicin comparable to that used here [
15] showed the presence of vascular breakdown and hemorrhage 48 hours after treatment. In contrast, the results summarized here were obtained using tumors harvested one week after the final treatment; thus the data here may reflect late effects on tumor vasculature. Further, 9L is a gliosarcoma cell line which exhibits a slower doubling time (34.9 hrs [
50]) than the U251MG glioblastoma cell line (20.9 hrs; data not shown) used in this study. The resulting 9L tumors are also histological distinct [
50] when compared to the U251MG model. These differences will likely impact how tumors respond to agents capable of promoting vascular normalization. Studies assessing how vascular functions change in relationship to tumor growth rate are currently being completed.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
MV carried out all parts of the experimental manipulations, data analysis and draft of manuscript. DS and DM were involved in the implantation of s.c. and orthotopic tumors and monitoring of the animals. MA and DW were involved in the development of Irinophore C™ formulation. AY and PK were part of MRI-DCE data acquisition and analysis. MBB and DTY were involved in the conception of the study, participated in its design and helped to draft the manuscript. All authors read and approved the final manuscript.