Spatial and temporal mapping of heterogeneity in liposome uptake and microvascular distribution in an orthotopic tumor xenograft model Sandra N. Ekdawi, James M.P. Stewart, Michael Dunne, Shawn Stapleton, Nicholas Mitsakakis, Yannan N. Dou, David A. Jaffray, Christine Allen PII: DOI: Reference:
S0168-3659(15)00224-2 doi: 10.1016/j.jconrel.2015.04.006 COREL 7627
To appear in:
Journal of Controlled Release
Received date: Revised date: Accepted date:
26 December 2014 21 March 2015 4 April 2015
Please cite this article as: Sandra N. Ekdawi, James M.P. Stewart, Michael Dunne, Shawn Stapleton, Nicholas Mitsakakis, Yannan N. Dou, David A. Jaffray, Christine Allen, Spatial and temporal mapping of heterogeneity in liposome uptake and microvascular distribution in an orthotopic tumor xenograft model, Journal of Controlled Release (2015), doi: 10.1016/j.jconrel.2015.04.006
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT Spatial and temporal mapping of heterogeneity in liposome uptake and microvascular distribution in an orthotopic tumor xenograft model
PT
Sandra N. Ekdawi1, James M. P. Stewart2, Michael Dunne1, Shawn Stapleton3,5, Nicholas Mitsakakis1, Yannan N. Dou1, David A. Jaffray2-7 and Christine Allen1* 1
Leslie Dan Faculty of Pharmacy, University of Toronto, Toronto, Ontario M5S 3M2, Canada Institute of Biomaterials and Biomedical Engineering, University of Toronto, Toronto, Ontario M5S 3G9, Canada 3 Department of Medical Biophysics, University of Toronto, Toronto, Ontario M5G 1L7, Canada 4 Department of Radiation Oncology, University of Toronto, Toronto, Ontario M5S 3E2, Canada 5 Radiation Medicine Program, Princess Margaret Cancer Centre, University Health Network, Toronto, Ontario M5G 2M9, Canada 6 Ontario Cancer Institute, Toronto, Ontario M5G 2M9, Canada 7 Techna Institute, University Health Network, Toronto, Ontario M5G 1P5, Canada
NU
SC
RI
2
AC CE P
Abstract
TE
D
MA
*Corresponding author: Christine Allen, PhD Leslie Dan Faculty of Pharmacy, University of Toronto 144 College Street, Toronto, Ontario, M5S 3M2, Canada Tel.: +1 416 946 8594 Fax: +1 416 978 8511 e-mail: [email protected]
Existing paradigms in nano-based drug delivery are currently being challenged. Assessment of bulk tumor accumulation has been routinely considered an indicative measure of nanomedicine potency. However, it is now recognized that the intratumoral distribution of nanomedicines also impacts their therapeutic effect. At this time, our understanding of the relationship between the bulk (i.e., macro-) tumor accumulation of nanocarriers and their intratumoral (i.e., micro-) distribution remains limited. Liposome-based drug formulations, in particular, suffer from diminished efficacy in vivo as a result of transport-limiting properties, combined with the heterogeneous nature of the tumor microenvironment. In this report, we perform a quantitative image-based assessment of macro- and microdistribution of liposomes. Multi-scalar assessment of liposome distribution was enabled by a stable formulation which co-encapsulates an iodinated contrast agent and a near-infrared fluorescence probe, for computed tomography (CT) and optical microscopy, respectively. Spatio-temporal quantification of tumor uptake in orthotopic xenografts was performed using CT at the bulk tissue level, and within defined sub-volumes of the tumor (i.e., rim, periphery and core). Tumor penetration and relative distribution of liposomes were assessed by fluorescence microscopy of whole tumor sections. Microdistribution analysis of whole tumor images exposed a heterogeneous distribution of both liposomes and tumor vasculature. Highest levels of liposome uptake were achieved and maintained in the wellvascularized tumor rim over the study period, corresponding to a positive correlation between liposome and microvascular density. Tumor penetration of liposomes was found to be timedependent in all regions of the tumor however independent of location in the tumor. Importantly, a multi-scalar comparison of liposome distribution reveals that macro-accumulation
1
ACCEPTED MANUSCRIPT in tissues (e.g., blood, whole tumor) may not reflect micro-accumulation levels present within specific regions of the tumor as a function of time. Keywords
PT
Nanomedicine; liposome; intratumoral distribution; tumor accumulation; computed tomography; optical microscopy
RI
1. Introduction
TE
D
MA
NU
SC
Tumor resistance to liposome-based therapy has been linked to heterogeneous tissue distribution and limited penetration of both nanocarrier and drug. Increasing reports of the intratumoral distribution of small-molecule [1-5], macromolecular [6], and nanoparticle-based agents [7-10] have yielded insight into the impact of the physico-chemical properties of the drug delivery system as well as that of the tumor microenvironment [11] on anti-tumor efficacy. Specifically, the fate of nano-based agents at the tumor site has been examined in relation to select pathophysiological properties of tumors deemed critical to the success of nanomedicines such as the distribution of the tumor vascular network [11], vascular density [12, 13] and permeability [14, 15], as well as the composition and density of scaffold proteins of the extracellular matrix [16, 17]. Such studies have significantly contributed to our understanding of the underlying barriers hindering the homogeneous distribution of nanomedicines within tumors. In turn, strategic exploitation of tumor-specific properties has been achieved through physical and pharmacological modulators, enabling enhanced delivery of drug, and/or superior anti-tumor efficacy [7, 18-21].
AC CE P
Despite an increased focus on the microscopic distribution of nanoparticles and/or their cargo in tumors, the relationship between macro- (i.e., bulk) and microdistribution of advanced drug delivery systems remains to be elucidated. This is particularly important given the chronic overreliance on the evaluation of bulk tumor accumulation of nanosystems as indicative of their in vivo performance. The ability of nanosystems to accumulate preferentially at the tumor site is attributed to the hyperpermeability of the tumor vasculature and impaired lymphatic clearance system; a phenomenon defined as the enhanced permeability and retention (EPR) effect [22, 23]. Recognized over the past three decades as a universal trait of solid tumors, the EPR effect has recently become known as somewhat of a “moving target” [24]. Indeed, inherent pathophysiological variability, as well as the impact of therapy and/or modulators, influence the status of the EPR effect, both spatially and temporally, in a given tumor and for a given therapeutic [25, 26]. The ensuing effect on both macro- and microdistribution of nanomedicines, and in turn on their anti-tumor efficacy, remains poorly characterized Similarly, further investigation into the relationship between the microdistribution of nanoparticles and tumor microenvironmental parameters, such as microvessel density (MVD), is pertinent. The tumor microenvironment (TME) has indeed been implicated in the resistance of lesions to both conventional and nanoparticle-based therapy. In particular, aberrant tumor vascular structure and function, solid stress, and interstitial hypertension [27] exacerbate the heterogeneous tumor distribution of delivered therapeutics, resulting in their limited penetration and/or anti-tumor activity [28-30]. Variability in tumor properties has been reported across tumor types, among tumors of the same type as well as within the same tumor [31-33]. As such, the heterogeneity itself in the status of such properties calls for their spatio-temporal characterization and subsequent relation to the delivery, and ultimately the efficacy, of a specific nanomedicine.
2
ACCEPTED MANUSCRIPT
AC CE P
TE
D
MA
NU
SC
RI
PT
We investigate the relationship between the tumor macro- and microdistribution of liposomes, as well as that between their microdistribution and properties of the TME. Such characterization is expected to yield a methodological platform which may further enable a greater understanding of macro- and microscopic parameters as potential determinants of the efficacy of nanomedicines. Hence, an imageable and stable liposome formulation is required which can be detected at both levels of resolution over the course of the experiment. We have therefore built upon our previous studies which have employed computed tomography (CT) as a quantitative imaging modality to assess the macrodistribution of liposomes [34, 35], and optical microscopy as a means to assess the tumor penetration of block copolymer micelles [8, 9]. As such, tissue deposition, distribution and penetration can be measured using the same liposome formulation via complementary contrast agents and corresponding imaging modalities. CT permits quantification and sub-mm resolution of liposomes while fluorescence microscopy enables visualization of liposome distribution at the sub-μm level relative to select factors of the TME. Overall, this study presents a framework to analyze the macro- and microdistribution of nanosystems in vivo. Specifically, spatio-temporal characterization of the intratumoral distribution of liposomes and tumor properties is performed quantitatively. Beyond bulk tumor characterization, microdistribution measurements provide site-specific information, revealing differences in inter-region liposome accumulation and microvascular density. Such differences may reveal trends. This is shown in the relationship found between liposome concentration and MVD, highlighting the key role that the tumor vasculature plays in defining the spatio-temporal tumor distribution of nanoparticles. Tumor penetration of liposomes is also characterized as a function of tumor region and time, revealing the contribution of both variables in determining liposome transport. Importantly, we show that systemic (i.e., plasma) and bulk tumor accumulation levels of liposomes are not necessarily predictive of the levels present within specific regions of the tumor. Such findings are expected to guide the evaluation and successful implementation of nanomedicines. 2. Materials and methods 2.1. Materials
1,2-dipalmitoyl-sn-glycero-3-phosphocholine (DPPC) and N-(carbonyl-methoxypolyethylene glycol-2000)-1,2-distearoyl-sn-glycero-3-phosphoethanolamine (mPEG2000-DSPE) were purchased from Genzyme Pharmaceuticals (Cambridge, MA). Cholesterol was obtained from Northern Lipids (Burnaby, BC, Canada). The small-molecule iodinated CT agent, iohexol (IOX), was obtained as a 350 mg I / mL solution of Omnipaque® from GE Healthcare (Mississauga, Canada). The near-infrared (NIR) fluorescent dye, Genhance™ 680 (GH680), was purchased from PerkinElmer (Woodbridge, Canada). 2.2. Liposome preparation The liposomes used in this work were developed from previously published protocols employing IOX as a contrast agent [35, 36], to co-encapsulate the NIR fluorescent dye GH680. Briefly, lipids were dissolved in anhydrous ethanol at 72°C in a molar ratio of 55:40:5 DPPC:CH:PEG2000-DSPE. Following evaporation of the ethanol, an aqueous solution of IOX (350 mg I /mL) and GH680 (1.11 mg/mL) was added to yield a lipid concentration of 100 mM. The solution was kept at 72°C for 4 hours with frequent vortexing, followed by stirring at room temperature overnight. Unilamellar liposomes were obtained by 5 extrusion cycles through two
3
ACCEPTED MANUSCRIPT
SC
RI
PT
stacked 200 nm pore size Track-Etch polycarbonate membranes (Whatman Inc., Clifton, NJ) at a pressure of 250 psi, followed by 10 cycles through two stacked 80 nm membranes at 400 psi using a 10 mL Lipex Extruder (Northern Lipids Inc., Burnaby, Canada). Between extrusion through the 200 nm and 80 nm membranes, the liposomes were allowed to stir overnight at room temperature. Unencapsulated IOX and GH680 were removed by dialysis (MWCO 50 kDa, Spectrum Labs, Rancho Dominguez, CA) against a minimum 250-fold volume excess of 0.02 mM HEPES-buffered saline solution (HBS, pH 7.4) over a period of 4 days. The liposome solution was then concentrated using a tangential flow column (MidGee ultrafiltration cartridge, 750,000 NMWC pore size, GE Healthcare, Mississauga, ON, Canada) and peristaltic pump (Watson Marlow Inc., Wilmington, MA) to a final lipid concentration of approximately 200 mM, and an iodine concentration of approximately 70 mg/mL.
NU
2.3. Liposome characterization
AC CE P
TE
D
MA
The hydrodynamic diameter, zeta potential, and contrast agent loading of the liposomes were measured. Briefly, the liposomes were diluted 250x in deionized water and their mean intensitybased and number-based diameters were recorded, as well as zeta potential, using a Malvern zetasizer instrument (Malvern, UK). The concentrations of encapsulated IOX and GH680 were determined by HPLC (Agilent Technologies 1200 Series). Specifically, a 10-fold volume excess of methanol, followed by thorough vortexing, was used to rupture the liposomes and a further 100-fold dilution in 90% methanol was performed in order to simultaneously detect both imaging agents within their respective linear ranges. The samples were injected onto an XTerra® MS C18 reverse-phase column (5 μm particle size, 4.6 mm x 250 mm dimensions; Waters Ltd., Mississauga, Canada) and eluted using a mobile phase consisting of a gradient of triethylammomium acetate (TEAA) buffer (pH 5.2) and methanol at a flow rate of 0.8 mL/min. IOX was detected at 245 nm (Waters 2487 Dual λ Absorbance Detector), while GH680 was detected using a Series 200a Fluorescence Detector (PerkinElmer, Woodbridge, Canada). The concentrations of encapsulated IOX and GH680 were determined from standard curves for each agent and are reported as the mean ± standard deviation of three independent liposome preparations. All measurements for each liposome batch were performed in triplicate. 2.4. Liposome pharmacokinetics and in vivo stability Severe combined immunodeficient (SCID) female mice (8-9 weeks old) were injected intravenously (iv) via the tail vein with the liposomes (1.22 mg lipid/ g mouse; 0.7 mg I per g mouse; 2 μg GH680 per g mouse). At 0.17, 2, 4, 8, 18, 24, 48, 72, 96 and 120 h post-injection (hpi), blood samples were collected in a heparinized tube following puncture of the saphenous vein. Similarly, a control group of mice received the same dose of free contrast agents in solution (i.e., not liposome-encapsulated). Immediately following blood collection, plasma was isolated via centrifugation at 2320 g for 5 minutes at 4°C and stored at -20°C until subsequent HPLC analysis. Extraction of IOX and GH680 from plasma was performed using a 10-fold volume excess of ice-cold methanol, followed by centrifugation at 20000 g for 30 minutes. The resulting supernatant was aliquoted for direct HPLC analysis under the same conditions described for liposome loading analysis. Plasma concentration versus time curves of the contrast agents were fit using a two-compartment model weighted by 1/SD2 in GraphPad Prism v 5.03. For each contrast agent, the distribution half-life (t1/2ɑ ), elimination half-life (t1/2β), volume of distribution (Vd), plasma area under the curve (AUC0-120h) and clearance (CL) were calculated and presented 4
ACCEPTED MANUSCRIPT as mean ± standard deviation of n=8 (liposome-) injected mice. A Pearson product-moment correlation was performed as a measure of retention of both contrast agents within the liposomes.
PT
2.5. Animal imaging and CT analysis
AC CE P
TE
D
MA
NU
SC
RI
All animal procedures were approved by the University Health Network Animal Care Committee. ME180 cervical tumors were grown in SCID mice by orthotopic implantation as described previously [37, 38]. Once tumors reached approximately 400-600 mm3 in volume, as determined by MRI, mice were iv administered liposomes at the same dose as the pharmacokinetics study (i.e., 1.22 mg lipid/ g mouse). The animals were subsequently scanned by CT (GE Locus Ultra MicroCT, GE Healthcare, Waunakee, WI) at 0.17, 2, 4, 8, 18, 24, 48, 72, 96, and 120 hpi, with pre-injection scans recorded for all mice as baseline. As described previously [35], CT scans of whole mouse were acquired at 80 kVp and 50 mA within a defined field of view. Scans were then reconstructed with an isotropic voxel size of ~153 μm and subsequently, three-dimensional (3D) volumes of interest (VOI) were generated for tumor and normal tissues (i.e., kidneys, liver and spleen) at each time point. The VOIs were created using a semi-automatic approach that combined manual contouring and threshold refinement (MicroView, GE Healthcare, Waunakee, WI). Mean CT number (expressed in Hounsfield units, HU) resulting from liposome-mediated contrast enhancement was determined for the whole tumor, kidneys, liver and spleen at all time points defined. These data were converted to %ID/cm3 tissue following a linear relationship between iodine concentration and HU as reported previously [35, 39]. Intratumoral analysis of liposome accumulation employed a custom erosion algorithm implemented in MATLAB (Mathworks, Natick, MA) in order to divide each tumor VOI into 3 concentric sub-volumes that encompassed the tumor rim, periphery and core. The width of the rim was defined as 10% of the radius of the VOI (where the radius was determined from a sphere whose volume is equal to the tumor VOI). The tumor concentration of the liposomes was determined in each sub-volume following the same method outlined above for determining bulk accumulation as %ID/cm3. The plasma concentration of the liposomes was determined in the descending aorta and adjusted for the arterial hematocrit ( ). The average plasma volume fraction of each tumor was estimated using early time point imaging of the liposomes (i.e., at 10 min post-injection). At an early time point, the liposomes are assumed to be predominantly intravascular. The plasma volume fraction was determined by taking the ratio of average liposome iodine concentration measured in each tumor sub-volume to that in plasma 10 min post-injection. The plasma volume fraction was used to subtract the contribution of the vascular liposomes from the measured tumor concentration of liposomes, providing an estimate of the extravascular concentration of liposomes in each tumor sub-volume. 2.6. Histology and fluorescence microscopy Following iv liposome administration, ME180 tumors were excised at 2, 18, 48, and 120 hpi for analysis of liposome microdistribution. Following animal sacrifice, tumors were excised and placed in cryomolds (Sakura Finetek, Torrance, CA) using OCT compound (Tissue-Tek, Sakura Finetek, Torrance, CA) as embedding medium and immediately frozen in liquid nitrogen. Tumor blocks were subsequently stored at -80°C until histological processing. Tumor sections 5 μmthick were prepared using a cryostat and mounted on glass slides. Sections were then imaged
5
ACCEPTED MANUSCRIPT
RI
PT
using an Olympus BX50 upright fluorescence microscope (Olympus, PA) at 10x magnification (Olympus UPlanSApo 10x/0.40) using a Semrock Quad Sedat filter set. Liposome signal was first captured from unstained tumor sections in the near-infrared range (650/684 nm). Subsequently, the tumor sections were stained and re-imaged to visualize blood vessels (antiCD31; 560/607 nm). The microscope was equipped with an EXFO fluorescence illumination source for which illumination power was monitored over the course of the studies. Images were acquired using a Photometrics CoolSnap HQ2 CCD camera and a motorized stage. Fixed exposure and contrast settings were applied in all tumor images.
SC
2.7. Fluorescence image pre-processing
AC CE P
TE
D
MA
NU
Images acquired from the microscope were comprised of individual tiles of whole tumor sections that were stitched together to generate a whole tumor image (MetaMorph®, Molecular Devices, Sunnyvale, CA) for the unstained section (i.e., liposomes) and stained section that was subsequently imaged (i.e., CD31-positive blood vessels) (Figure 1A). All images obtained for a given tumor were cropped to the same dimensions and aligned (i.e., registered without morphological changes) using ImagePro PLUS software (Media Cybernetics, Rockville, MD). Individual tumor masks were then created to exclude non-tumor tissue and artifact resulting from sectioning and staining (e.g., tissue folds, tears) using FIJI software [40] and applied to all images of a given tumor section. Resulting masked images were preserved as 16-bit grayscale and employed as such in subsequent image analysis (Figure 1B).
6
ACCEPTED MANUSCRIPT
SC
RI
PT
Figure 1. Representation of fluorescence image analysis process. (A) Liposome (red) and blood vessel (BV, green) signals are acquired separately, requiring (B) pre-processing to align both images. (C) Images are then processed via a customized MATLAB algorithm which permits segmentation of liposome and BV signals. BV signal (binary) produces a distance map illustrated in grayscale whereby black represents pixels most proximal to the nearest BV and white represents those most distal, according to a gradient. Liposome signal intensity (SI, continuous) is segmented according to a global threshold and represented as mean SI as a function of distance to nearest BV, measured in bins of 1 μm, within the tumor rim (r), periphery (p) and core (c). Scale bars in whole tumor images represent 1 mm while those in the magnifications represent 200 μm.
NU
2.8. Fluorescence image analysis
AC CE P
TE
D
MA
A computational methodology for analyzing the distance of liposomes from blood vessels within whole tumor sections has been developed based on previous work with block copolymer micelles [8, 9]. In our study, we generated a distance map based on the image of CD31-stained and thresholded blood vessels within the whole tumor section, and subsequently identified liposome signal within three distinct, concentric regions of the corresponding tumor section (i.e., rim, periphery and core). The mean liposome signal intensity versus distance to nearest vessel was then extracted and plotted for each of the three regions (Figure 1C). The motivation for such spatial partitioning in our analysis emerged from the observation of a characteristic “rim effect” found in several different tumor models, revealing the region-specific accumulation of liposomes similar in composition and size to those in the present study [27, 34], as well as reports of a radial pressure gradient typically observed in solid tumors [41, 42]. The width of the rim was defined as 10% of the maximum radius of the tumor, while the periphery and core were divided into areas of equal width. The tissue background levels in each liposome image were measured and subtracted in order to normalize background levels across the data set. All liposome images were analyzed using a global threshold value in order to systematically segment liposome-positive signal over time. Liposome concentration was approximated using the mean fluorescent signal intensity (i.e., total fluorescent signal per unit area) and was measured as a function of distance to nearest blood vessel. Vascular images underwent automated segmentation using Otsu’s method to differentiate between CD31-positive and negative regions, and were subsequently masked to generate a distance map. A characteristic penetration length (CPL) was defined as the distance from the tumor vascular endothelium within which lies 50% of the cumulative mean liposome signal intensity. CD31-positive objects less than 10 μm in size were excluded from the analysis based on an estimated minimum capillary diameter [43]. Microvessel density (MVD) within each region, as well as in the tumor as a whole, was determined as CD31positive area over total tumor area. Images displayed in this paper were adjusted for brightness and contrast for presentation purposes only. 2.9. Statistical analysis Comparisons between two groups were conducted using Student’s t-test; namely, for in vivo stability of the liposome construct (i.e., between co-encapsulated contrast agents). Pearson correlation was employed for further analysis of in vivo liposome stability, as well as to assess
7
ACCEPTED MANUSCRIPT
PT
the relationship between tumor MVD and liposome accumulation. Linear regression analysis was used to evaluate liposome penetration over time. Measures of tumor accumulation, distribution, and penetration were assessed over time and/or within regions by one-way ANOVA followed by Tukey’s HSD test. Multiple testing corrections were not applied, and statistical significance was attained when p
S0168-3659(15)00224-2 doi: 10.1016/j.jconrel.2015.04.006 COREL 7627
To appear in:
Journal of Controlled Release
Received date: Revised date: Accepted date:
26 December 2014 21 March 2015 4 April 2015
Please cite this article as: Sandra N. Ekdawi, James M.P. Stewart, Michael Dunne, Shawn Stapleton, Nicholas Mitsakakis, Yannan N. Dou, David A. Jaffray, Christine Allen, Spatial and temporal mapping of heterogeneity in liposome uptake and microvascular distribution in an orthotopic tumor xenograft model, Journal of Controlled Release (2015), doi: 10.1016/j.jconrel.2015.04.006
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT Spatial and temporal mapping of heterogeneity in liposome uptake and microvascular distribution in an orthotopic tumor xenograft model
PT
Sandra N. Ekdawi1, James M. P. Stewart2, Michael Dunne1, Shawn Stapleton3,5, Nicholas Mitsakakis1, Yannan N. Dou1, David A. Jaffray2-7 and Christine Allen1* 1
Leslie Dan Faculty of Pharmacy, University of Toronto, Toronto, Ontario M5S 3M2, Canada Institute of Biomaterials and Biomedical Engineering, University of Toronto, Toronto, Ontario M5S 3G9, Canada 3 Department of Medical Biophysics, University of Toronto, Toronto, Ontario M5G 1L7, Canada 4 Department of Radiation Oncology, University of Toronto, Toronto, Ontario M5S 3E2, Canada 5 Radiation Medicine Program, Princess Margaret Cancer Centre, University Health Network, Toronto, Ontario M5G 2M9, Canada 6 Ontario Cancer Institute, Toronto, Ontario M5G 2M9, Canada 7 Techna Institute, University Health Network, Toronto, Ontario M5G 1P5, Canada
NU
SC
RI
2
AC CE P
Abstract
TE
D
MA
*Corresponding author: Christine Allen, PhD Leslie Dan Faculty of Pharmacy, University of Toronto 144 College Street, Toronto, Ontario, M5S 3M2, Canada Tel.: +1 416 946 8594 Fax: +1 416 978 8511 e-mail: [email protected]
Existing paradigms in nano-based drug delivery are currently being challenged. Assessment of bulk tumor accumulation has been routinely considered an indicative measure of nanomedicine potency. However, it is now recognized that the intratumoral distribution of nanomedicines also impacts their therapeutic effect. At this time, our understanding of the relationship between the bulk (i.e., macro-) tumor accumulation of nanocarriers and their intratumoral (i.e., micro-) distribution remains limited. Liposome-based drug formulations, in particular, suffer from diminished efficacy in vivo as a result of transport-limiting properties, combined with the heterogeneous nature of the tumor microenvironment. In this report, we perform a quantitative image-based assessment of macro- and microdistribution of liposomes. Multi-scalar assessment of liposome distribution was enabled by a stable formulation which co-encapsulates an iodinated contrast agent and a near-infrared fluorescence probe, for computed tomography (CT) and optical microscopy, respectively. Spatio-temporal quantification of tumor uptake in orthotopic xenografts was performed using CT at the bulk tissue level, and within defined sub-volumes of the tumor (i.e., rim, periphery and core). Tumor penetration and relative distribution of liposomes were assessed by fluorescence microscopy of whole tumor sections. Microdistribution analysis of whole tumor images exposed a heterogeneous distribution of both liposomes and tumor vasculature. Highest levels of liposome uptake were achieved and maintained in the wellvascularized tumor rim over the study period, corresponding to a positive correlation between liposome and microvascular density. Tumor penetration of liposomes was found to be timedependent in all regions of the tumor however independent of location in the tumor. Importantly, a multi-scalar comparison of liposome distribution reveals that macro-accumulation
1
ACCEPTED MANUSCRIPT in tissues (e.g., blood, whole tumor) may not reflect micro-accumulation levels present within specific regions of the tumor as a function of time. Keywords
PT
Nanomedicine; liposome; intratumoral distribution; tumor accumulation; computed tomography; optical microscopy
RI
1. Introduction
TE
D
MA
NU
SC
Tumor resistance to liposome-based therapy has been linked to heterogeneous tissue distribution and limited penetration of both nanocarrier and drug. Increasing reports of the intratumoral distribution of small-molecule [1-5], macromolecular [6], and nanoparticle-based agents [7-10] have yielded insight into the impact of the physico-chemical properties of the drug delivery system as well as that of the tumor microenvironment [11] on anti-tumor efficacy. Specifically, the fate of nano-based agents at the tumor site has been examined in relation to select pathophysiological properties of tumors deemed critical to the success of nanomedicines such as the distribution of the tumor vascular network [11], vascular density [12, 13] and permeability [14, 15], as well as the composition and density of scaffold proteins of the extracellular matrix [16, 17]. Such studies have significantly contributed to our understanding of the underlying barriers hindering the homogeneous distribution of nanomedicines within tumors. In turn, strategic exploitation of tumor-specific properties has been achieved through physical and pharmacological modulators, enabling enhanced delivery of drug, and/or superior anti-tumor efficacy [7, 18-21].
AC CE P
Despite an increased focus on the microscopic distribution of nanoparticles and/or their cargo in tumors, the relationship between macro- (i.e., bulk) and microdistribution of advanced drug delivery systems remains to be elucidated. This is particularly important given the chronic overreliance on the evaluation of bulk tumor accumulation of nanosystems as indicative of their in vivo performance. The ability of nanosystems to accumulate preferentially at the tumor site is attributed to the hyperpermeability of the tumor vasculature and impaired lymphatic clearance system; a phenomenon defined as the enhanced permeability and retention (EPR) effect [22, 23]. Recognized over the past three decades as a universal trait of solid tumors, the EPR effect has recently become known as somewhat of a “moving target” [24]. Indeed, inherent pathophysiological variability, as well as the impact of therapy and/or modulators, influence the status of the EPR effect, both spatially and temporally, in a given tumor and for a given therapeutic [25, 26]. The ensuing effect on both macro- and microdistribution of nanomedicines, and in turn on their anti-tumor efficacy, remains poorly characterized Similarly, further investigation into the relationship between the microdistribution of nanoparticles and tumor microenvironmental parameters, such as microvessel density (MVD), is pertinent. The tumor microenvironment (TME) has indeed been implicated in the resistance of lesions to both conventional and nanoparticle-based therapy. In particular, aberrant tumor vascular structure and function, solid stress, and interstitial hypertension [27] exacerbate the heterogeneous tumor distribution of delivered therapeutics, resulting in their limited penetration and/or anti-tumor activity [28-30]. Variability in tumor properties has been reported across tumor types, among tumors of the same type as well as within the same tumor [31-33]. As such, the heterogeneity itself in the status of such properties calls for their spatio-temporal characterization and subsequent relation to the delivery, and ultimately the efficacy, of a specific nanomedicine.
2
ACCEPTED MANUSCRIPT
AC CE P
TE
D
MA
NU
SC
RI
PT
We investigate the relationship between the tumor macro- and microdistribution of liposomes, as well as that between their microdistribution and properties of the TME. Such characterization is expected to yield a methodological platform which may further enable a greater understanding of macro- and microscopic parameters as potential determinants of the efficacy of nanomedicines. Hence, an imageable and stable liposome formulation is required which can be detected at both levels of resolution over the course of the experiment. We have therefore built upon our previous studies which have employed computed tomography (CT) as a quantitative imaging modality to assess the macrodistribution of liposomes [34, 35], and optical microscopy as a means to assess the tumor penetration of block copolymer micelles [8, 9]. As such, tissue deposition, distribution and penetration can be measured using the same liposome formulation via complementary contrast agents and corresponding imaging modalities. CT permits quantification and sub-mm resolution of liposomes while fluorescence microscopy enables visualization of liposome distribution at the sub-μm level relative to select factors of the TME. Overall, this study presents a framework to analyze the macro- and microdistribution of nanosystems in vivo. Specifically, spatio-temporal characterization of the intratumoral distribution of liposomes and tumor properties is performed quantitatively. Beyond bulk tumor characterization, microdistribution measurements provide site-specific information, revealing differences in inter-region liposome accumulation and microvascular density. Such differences may reveal trends. This is shown in the relationship found between liposome concentration and MVD, highlighting the key role that the tumor vasculature plays in defining the spatio-temporal tumor distribution of nanoparticles. Tumor penetration of liposomes is also characterized as a function of tumor region and time, revealing the contribution of both variables in determining liposome transport. Importantly, we show that systemic (i.e., plasma) and bulk tumor accumulation levels of liposomes are not necessarily predictive of the levels present within specific regions of the tumor. Such findings are expected to guide the evaluation and successful implementation of nanomedicines. 2. Materials and methods 2.1. Materials
1,2-dipalmitoyl-sn-glycero-3-phosphocholine (DPPC) and N-(carbonyl-methoxypolyethylene glycol-2000)-1,2-distearoyl-sn-glycero-3-phosphoethanolamine (mPEG2000-DSPE) were purchased from Genzyme Pharmaceuticals (Cambridge, MA). Cholesterol was obtained from Northern Lipids (Burnaby, BC, Canada). The small-molecule iodinated CT agent, iohexol (IOX), was obtained as a 350 mg I / mL solution of Omnipaque® from GE Healthcare (Mississauga, Canada). The near-infrared (NIR) fluorescent dye, Genhance™ 680 (GH680), was purchased from PerkinElmer (Woodbridge, Canada). 2.2. Liposome preparation The liposomes used in this work were developed from previously published protocols employing IOX as a contrast agent [35, 36], to co-encapsulate the NIR fluorescent dye GH680. Briefly, lipids were dissolved in anhydrous ethanol at 72°C in a molar ratio of 55:40:5 DPPC:CH:PEG2000-DSPE. Following evaporation of the ethanol, an aqueous solution of IOX (350 mg I /mL) and GH680 (1.11 mg/mL) was added to yield a lipid concentration of 100 mM. The solution was kept at 72°C for 4 hours with frequent vortexing, followed by stirring at room temperature overnight. Unilamellar liposomes were obtained by 5 extrusion cycles through two
3
ACCEPTED MANUSCRIPT
SC
RI
PT
stacked 200 nm pore size Track-Etch polycarbonate membranes (Whatman Inc., Clifton, NJ) at a pressure of 250 psi, followed by 10 cycles through two stacked 80 nm membranes at 400 psi using a 10 mL Lipex Extruder (Northern Lipids Inc., Burnaby, Canada). Between extrusion through the 200 nm and 80 nm membranes, the liposomes were allowed to stir overnight at room temperature. Unencapsulated IOX and GH680 were removed by dialysis (MWCO 50 kDa, Spectrum Labs, Rancho Dominguez, CA) against a minimum 250-fold volume excess of 0.02 mM HEPES-buffered saline solution (HBS, pH 7.4) over a period of 4 days. The liposome solution was then concentrated using a tangential flow column (MidGee ultrafiltration cartridge, 750,000 NMWC pore size, GE Healthcare, Mississauga, ON, Canada) and peristaltic pump (Watson Marlow Inc., Wilmington, MA) to a final lipid concentration of approximately 200 mM, and an iodine concentration of approximately 70 mg/mL.
NU
2.3. Liposome characterization
AC CE P
TE
D
MA
The hydrodynamic diameter, zeta potential, and contrast agent loading of the liposomes were measured. Briefly, the liposomes were diluted 250x in deionized water and their mean intensitybased and number-based diameters were recorded, as well as zeta potential, using a Malvern zetasizer instrument (Malvern, UK). The concentrations of encapsulated IOX and GH680 were determined by HPLC (Agilent Technologies 1200 Series). Specifically, a 10-fold volume excess of methanol, followed by thorough vortexing, was used to rupture the liposomes and a further 100-fold dilution in 90% methanol was performed in order to simultaneously detect both imaging agents within their respective linear ranges. The samples were injected onto an XTerra® MS C18 reverse-phase column (5 μm particle size, 4.6 mm x 250 mm dimensions; Waters Ltd., Mississauga, Canada) and eluted using a mobile phase consisting of a gradient of triethylammomium acetate (TEAA) buffer (pH 5.2) and methanol at a flow rate of 0.8 mL/min. IOX was detected at 245 nm (Waters 2487 Dual λ Absorbance Detector), while GH680 was detected using a Series 200a Fluorescence Detector (PerkinElmer, Woodbridge, Canada). The concentrations of encapsulated IOX and GH680 were determined from standard curves for each agent and are reported as the mean ± standard deviation of three independent liposome preparations. All measurements for each liposome batch were performed in triplicate. 2.4. Liposome pharmacokinetics and in vivo stability Severe combined immunodeficient (SCID) female mice (8-9 weeks old) were injected intravenously (iv) via the tail vein with the liposomes (1.22 mg lipid/ g mouse; 0.7 mg I per g mouse; 2 μg GH680 per g mouse). At 0.17, 2, 4, 8, 18, 24, 48, 72, 96 and 120 h post-injection (hpi), blood samples were collected in a heparinized tube following puncture of the saphenous vein. Similarly, a control group of mice received the same dose of free contrast agents in solution (i.e., not liposome-encapsulated). Immediately following blood collection, plasma was isolated via centrifugation at 2320 g for 5 minutes at 4°C and stored at -20°C until subsequent HPLC analysis. Extraction of IOX and GH680 from plasma was performed using a 10-fold volume excess of ice-cold methanol, followed by centrifugation at 20000 g for 30 minutes. The resulting supernatant was aliquoted for direct HPLC analysis under the same conditions described for liposome loading analysis. Plasma concentration versus time curves of the contrast agents were fit using a two-compartment model weighted by 1/SD2 in GraphPad Prism v 5.03. For each contrast agent, the distribution half-life (t1/2ɑ ), elimination half-life (t1/2β), volume of distribution (Vd), plasma area under the curve (AUC0-120h) and clearance (CL) were calculated and presented 4
ACCEPTED MANUSCRIPT as mean ± standard deviation of n=8 (liposome-) injected mice. A Pearson product-moment correlation was performed as a measure of retention of both contrast agents within the liposomes.
PT
2.5. Animal imaging and CT analysis
AC CE P
TE
D
MA
NU
SC
RI
All animal procedures were approved by the University Health Network Animal Care Committee. ME180 cervical tumors were grown in SCID mice by orthotopic implantation as described previously [37, 38]. Once tumors reached approximately 400-600 mm3 in volume, as determined by MRI, mice were iv administered liposomes at the same dose as the pharmacokinetics study (i.e., 1.22 mg lipid/ g mouse). The animals were subsequently scanned by CT (GE Locus Ultra MicroCT, GE Healthcare, Waunakee, WI) at 0.17, 2, 4, 8, 18, 24, 48, 72, 96, and 120 hpi, with pre-injection scans recorded for all mice as baseline. As described previously [35], CT scans of whole mouse were acquired at 80 kVp and 50 mA within a defined field of view. Scans were then reconstructed with an isotropic voxel size of ~153 μm and subsequently, three-dimensional (3D) volumes of interest (VOI) were generated for tumor and normal tissues (i.e., kidneys, liver and spleen) at each time point. The VOIs were created using a semi-automatic approach that combined manual contouring and threshold refinement (MicroView, GE Healthcare, Waunakee, WI). Mean CT number (expressed in Hounsfield units, HU) resulting from liposome-mediated contrast enhancement was determined for the whole tumor, kidneys, liver and spleen at all time points defined. These data were converted to %ID/cm3 tissue following a linear relationship between iodine concentration and HU as reported previously [35, 39]. Intratumoral analysis of liposome accumulation employed a custom erosion algorithm implemented in MATLAB (Mathworks, Natick, MA) in order to divide each tumor VOI into 3 concentric sub-volumes that encompassed the tumor rim, periphery and core. The width of the rim was defined as 10% of the radius of the VOI (where the radius was determined from a sphere whose volume is equal to the tumor VOI). The tumor concentration of the liposomes was determined in each sub-volume following the same method outlined above for determining bulk accumulation as %ID/cm3. The plasma concentration of the liposomes was determined in the descending aorta and adjusted for the arterial hematocrit ( ). The average plasma volume fraction of each tumor was estimated using early time point imaging of the liposomes (i.e., at 10 min post-injection). At an early time point, the liposomes are assumed to be predominantly intravascular. The plasma volume fraction was determined by taking the ratio of average liposome iodine concentration measured in each tumor sub-volume to that in plasma 10 min post-injection. The plasma volume fraction was used to subtract the contribution of the vascular liposomes from the measured tumor concentration of liposomes, providing an estimate of the extravascular concentration of liposomes in each tumor sub-volume. 2.6. Histology and fluorescence microscopy Following iv liposome administration, ME180 tumors were excised at 2, 18, 48, and 120 hpi for analysis of liposome microdistribution. Following animal sacrifice, tumors were excised and placed in cryomolds (Sakura Finetek, Torrance, CA) using OCT compound (Tissue-Tek, Sakura Finetek, Torrance, CA) as embedding medium and immediately frozen in liquid nitrogen. Tumor blocks were subsequently stored at -80°C until histological processing. Tumor sections 5 μmthick were prepared using a cryostat and mounted on glass slides. Sections were then imaged
5
ACCEPTED MANUSCRIPT
RI
PT
using an Olympus BX50 upright fluorescence microscope (Olympus, PA) at 10x magnification (Olympus UPlanSApo 10x/0.40) using a Semrock Quad Sedat filter set. Liposome signal was first captured from unstained tumor sections in the near-infrared range (650/684 nm). Subsequently, the tumor sections were stained and re-imaged to visualize blood vessels (antiCD31; 560/607 nm). The microscope was equipped with an EXFO fluorescence illumination source for which illumination power was monitored over the course of the studies. Images were acquired using a Photometrics CoolSnap HQ2 CCD camera and a motorized stage. Fixed exposure and contrast settings were applied in all tumor images.
SC
2.7. Fluorescence image pre-processing
AC CE P
TE
D
MA
NU
Images acquired from the microscope were comprised of individual tiles of whole tumor sections that were stitched together to generate a whole tumor image (MetaMorph®, Molecular Devices, Sunnyvale, CA) for the unstained section (i.e., liposomes) and stained section that was subsequently imaged (i.e., CD31-positive blood vessels) (Figure 1A). All images obtained for a given tumor were cropped to the same dimensions and aligned (i.e., registered without morphological changes) using ImagePro PLUS software (Media Cybernetics, Rockville, MD). Individual tumor masks were then created to exclude non-tumor tissue and artifact resulting from sectioning and staining (e.g., tissue folds, tears) using FIJI software [40] and applied to all images of a given tumor section. Resulting masked images were preserved as 16-bit grayscale and employed as such in subsequent image analysis (Figure 1B).
6
ACCEPTED MANUSCRIPT
SC
RI
PT
Figure 1. Representation of fluorescence image analysis process. (A) Liposome (red) and blood vessel (BV, green) signals are acquired separately, requiring (B) pre-processing to align both images. (C) Images are then processed via a customized MATLAB algorithm which permits segmentation of liposome and BV signals. BV signal (binary) produces a distance map illustrated in grayscale whereby black represents pixels most proximal to the nearest BV and white represents those most distal, according to a gradient. Liposome signal intensity (SI, continuous) is segmented according to a global threshold and represented as mean SI as a function of distance to nearest BV, measured in bins of 1 μm, within the tumor rim (r), periphery (p) and core (c). Scale bars in whole tumor images represent 1 mm while those in the magnifications represent 200 μm.
NU
2.8. Fluorescence image analysis
AC CE P
TE
D
MA
A computational methodology for analyzing the distance of liposomes from blood vessels within whole tumor sections has been developed based on previous work with block copolymer micelles [8, 9]. In our study, we generated a distance map based on the image of CD31-stained and thresholded blood vessels within the whole tumor section, and subsequently identified liposome signal within three distinct, concentric regions of the corresponding tumor section (i.e., rim, periphery and core). The mean liposome signal intensity versus distance to nearest vessel was then extracted and plotted for each of the three regions (Figure 1C). The motivation for such spatial partitioning in our analysis emerged from the observation of a characteristic “rim effect” found in several different tumor models, revealing the region-specific accumulation of liposomes similar in composition and size to those in the present study [27, 34], as well as reports of a radial pressure gradient typically observed in solid tumors [41, 42]. The width of the rim was defined as 10% of the maximum radius of the tumor, while the periphery and core were divided into areas of equal width. The tissue background levels in each liposome image were measured and subtracted in order to normalize background levels across the data set. All liposome images were analyzed using a global threshold value in order to systematically segment liposome-positive signal over time. Liposome concentration was approximated using the mean fluorescent signal intensity (i.e., total fluorescent signal per unit area) and was measured as a function of distance to nearest blood vessel. Vascular images underwent automated segmentation using Otsu’s method to differentiate between CD31-positive and negative regions, and were subsequently masked to generate a distance map. A characteristic penetration length (CPL) was defined as the distance from the tumor vascular endothelium within which lies 50% of the cumulative mean liposome signal intensity. CD31-positive objects less than 10 μm in size were excluded from the analysis based on an estimated minimum capillary diameter [43]. Microvessel density (MVD) within each region, as well as in the tumor as a whole, was determined as CD31positive area over total tumor area. Images displayed in this paper were adjusted for brightness and contrast for presentation purposes only. 2.9. Statistical analysis Comparisons between two groups were conducted using Student’s t-test; namely, for in vivo stability of the liposome construct (i.e., between co-encapsulated contrast agents). Pearson correlation was employed for further analysis of in vivo liposome stability, as well as to assess
7
ACCEPTED MANUSCRIPT
PT
the relationship between tumor MVD and liposome accumulation. Linear regression analysis was used to evaluate liposome penetration over time. Measures of tumor accumulation, distribution, and penetration were assessed over time and/or within regions by one-way ANOVA followed by Tukey’s HSD test. Multiple testing corrections were not applied, and statistical significance was attained when p
