In solid tumors, rapid local intravascular release of anticancer agents, e.g., doxorubicin (DOX), from thermosensitive liposomes (TSLs) can be an option to overcome poor extravasation of drug nanocarriers. The driving force of DOX penetration is the drug concentration gradient between the vascular compartment and the tumor interstitium. In this feasibility study, we used fibered confocal fluorescence microscopy (FCFM) to monitor in real-time DOX penetration in the interstitium of a subcutaneous tumor after its intravascular release from TSLs, Thermodox®. Cell uptake kinetics of the released DOX was quantified, along with an in-depth assessment of released-DOX penetration using an evolution model. A subcutaneous rat R1 rhabdomyosarcoma xenograft was used. The rodent was positioned in a setup including a water bath, and FCFM identification of functional vessels in the tumor tissue was applied based on AngioSense. The tumor-bearing leg was immersed in the 43°C water for preheating, and TSLs were injected intravenously. Real-time monitoring of intratumoral (i.t.) DOX penetration could be performed, and it showed the progressing DOX wave front via its native fluorescence, labeling successively all cell nuclei. Cell uptake rates (1/k) of 3 minutes were found (
Local drug delivery strategies in oncology aim at increasing delivery of anticancer drugs in the tumor, while limiting their exposure in healthy tissues that induces toxic side effects, e.g., cardiotoxicity of the small molecule doxorubicin (DOX, relative molecular mass of 544 Da) [
The advent of temperature-sensitive liposomes (TSLs), with the first formulation described by Yatvin et al. [
Combination of LTSLs with an external heat source allowed local treatment, and Manzoor et al. introduced the concept of triggered, rapid intravascular release [
To conduct drug distribution studies at the tissue scale, dorsal skin-fold window chambers have been used
In this study, we tested the feasibility to use FCFM to monitor
All procedures were performed according to the ethical guidelines and were approved by the animal welfare committee of Utrecht University (DEC 2014.III.03.035, Utrecht, the Netherlands). WAG/Rij rats were purchased from Charles River (Cologne, Germany). They were maintained at room temperature with 12 h light cycle in individually ventilated isolation cages and were fed ad libitum. The rats were 12 weeks old at the beginning of the experiments, weighing 250 g. Under gaseous anesthesia (Aerrane, Baxter, Deerfield, IL), a skin incision of a few millimeters was performed at the hind leg. Subsequently, rat R1 rhabdomyosarcoma tumor pieces (1–3 mm3) were subcutaneously implanted in the hind leg using a trocar. When the tumor volume reached 1500
Lysothermosensitive liposomal formulation of DOX (Thermodox®-TSL) at 2 mg/mL was obtained from Celsion Corp (Lawrenceville, NJ, USA). These nanoparticles release their payload as a burst in the temperature range of 39.5°C to 42°C, i.e., less than 5% of release at 37°C, and more than 65% at 41°C, within about 30 sec (
Doxorubicin hydrochloride (Sigma-Aldrich, St-Louis, MO) (relative molecular mass: 580 Da), named “free DOX” in this study, was injected intravenously at 4 mg/kg.
An intravascular fluorescence label, AngioSense 680 EX, was purchased from Perkin Elmer (Waltham, MA, USA). AngioSense is a 70 kDa near-infrared labeled-fluorescent polymer (excitation/emission wavelengths: 670/690 nm), which allows imaging the blood pool during the whole imaging session.
Fluorescence images were acquired in real-time (8.5 Hz) for 20 minutes using a dual-band FCFM system (Cellvizio® dual-band, Mauna Kea Technologies, Paris, France). Native fluorescence of DOX was collected with the 488 nm excitation channel, henceforth referred to as “green channel,” and blood vessels via AngioSense with the 660 nm channel, referred to as “red channel”. Their spectral sensitivity is 500–630 nm and 680–800 nm, respectively. A 1.5 mm diameter FCFM microprobe (PF-2210, Mauna Kea Technologies) was used (Figure
Tumor access (a), setup (b), tip of the FCFM probe (c), and timeline of the experiment (d). After incision of the skin at the tumor location (a), the rat was positioned on a platform placed at the water surface of the water bath, with the hind leg immersed in the water set to 43°C (b).
A water bath (Memmert, Schwabach, Germany) was used to ensure a mild and homogeneous tumor heating (Figure
The animal was thermally isolated from the water bath using a piece of aluminium foil (Figure
Handling of the FCFM microprobe was facilitated by using a modular hose, holding it in fixed position in contact with the tumor tissue, such that a dynamic microscopy time series of a fixed tumor location could be obtained (Figure
The rats were anaesthetized with an i.p. injection of 75 mg/kg of ketamine (Narketan, Vetoquinol, ’s-Hertogenbosch, the Netherlands) and 0.25 mg/kg of dexmedetomidine (Dexdormitor, Orion Pharma, Mechelen, Belgium). Then, the jugular vein was catheterized, a 1 cm skin flap was created at the tumor level (Figure
Subsequently, the rats were positioned on the platform of the water bath, laying on the flank, with the tumor bearing leg immersed in the 43°C water. Temperature probes were then placed to monitor the temperature of the water bath and the body temperature of the animal. The tip of the FCFM probe was placed manually to make contact with the tumor tissue in the water, and the tumor was explored until tortuous tumor vessels were found. Then, stability of the FOV was verified by waiting one minute and by checking any motion of the tumor microvasculature in the image. The syringe containing Thermodox® was only then placed in the catheter, to avoid premature heating of the liposomes. At this point, the 20-minute real-time monitoring was started. After collecting the 10-second baseline with tissue autofluorescence, a 4 mg/kg bolus injection of TSLs, or free DOX, was administered intravenously in the jugular vein in around 40 s.
After completion of the dynamic sequence, exploration of the tumor surface was performed manually with the FCFM probe, and micrographs were acquired to evaluate the presence of DOX in the tumor microenvironment by means of its native fluorescence [
At the end of the session, the rats were sacrificed by an i.p. injection of 200 mg/kg of pentobarbital (Euthanimal 20%, Alfasan, Woerden, the Netherlands), and the blood, the urine if any, and the tumor were harvested.
Two treatment groups consisted of the “free DOX” group (
Tumors were harvested and fixed in the formol-acetic acid solution. Then, histological samples were embedded in paraffin, cut at approximately 5
Micrographs of DOX fluorescence in frozen-tissue sections were acquired using a Leica TCS SP8 X confocal fluorescence microscope, with a 10x magnification objective, a 504 nm excitation wavelength, and an emission filter of 540–680 nm.
DOX concentration in blood and urine samples was measured using Ultra Performance Liquid Chromatography (UPLC) with fluorescence and UV detection. An ACQUITY UPLC BEH C18 separation column was used (130 Ångström, size of 1.7 micron, 1.7 × 50 mm). The mobile phase consisted of an eluant with the following mixture: 75% reverse-osmosis water, 25% acetonitrile, and 1% perchloric acid. During separation, a column temperature of 50°C and a sample temperature of 25°C were set, with 3 minutes of run time. The fluorescence was measured with a 480 nm excitation/565 nm emission and a UV detection of 234 nm.
The real-time fluorescence image data obtained was processed offline using MATLAB® 2013 (The MathWorks, Natick, MA, USA). To increase the signal-to-noise ratio, the sequence was averaged temporally to an 8 s frame rate. Kinetic analysis of DOX penetration consisted of the 3 following independent but complementary sections.
Cell uptake kinetics was assessed using the dedicated parametric pipeline described in Derieppe et al. [
Quantitative evaluation of released-DOX penetration in the tumor interstitium was then performed by estimating the instantaneous apparent DOX transport (noted
The transport model of equation (
The data fidelity and the regularization terms are linked by the weighting factor
In order to mitigate the local impact of the nucleus fluorescence signal on DOX penetration in the interstitium, a spatial low-pass Butterworth filter was applied to each individual image before the resolution of equation (
The onset of fluorescence signal was defined when the maximum fluorescence signal in the current image exceeded at least 5% of the maximum signal of the sequence. Since the transport model of equation (
A principal component analysis (PCA) was applied subsequently on the time series of the computed 2-dimension penetration vector fields in order to find the spatial orthogonal basis; the resulting principal axis served to generate an adequate representation of the sequence of displacement fields. An averaged motion amplitude along each principal axis was then calculated for each individual displacement field and allowed calculating the temporal profile of the average motion amplitude along each principal axis.
Ultimately, the spatiotemporal distribution of released-DOX fluorescence signal intensity
In this model, the fluorescence signal intensity
Equation (
Since cell nuclei are nonmoving structures, the impact of their fluorescence signal was reduced using a spatial low-pass Butterworth filter (order 1) in each individual image before the numerical resolution of equation (
To evaluate the influence of mild hyperthermia on tumor tissue, histopathological analyses were carried out at the end of our imaging session. Tumors were visible as rounded neoplastic nodules present within the subcutaneous and muscle tissues and partially surrounded by thin fibrous capsules. All tumors (37°C vs 43°C) exhibited common histopathological characteristics. The proliferation areas were made up of poorly differentiated neoplastic cells within a fine vascular stroma. These cells were highly pleomorphic with one or more prominent nuclei. Some multinucleated cells were also observed in all tumors. The transition proliferative/necrotic areas are well-defined, with a loss in cell density and a clear fibrillary aspect that is characteristic of the extracellular matrix in the necrotic area (25–30% of all tumor nodules). Few multifocal infiltrations of neutrophils were observed in all tumors. No apparent tissue damage in any of the areas exposed to a 43°C local hyperthermia (Figure
In the red channel, staining of the blood pool by AngioSense allowed finding functional, characteristically tortuous, tumor microvasculature (Figures
Real-time monitoring of doxorubicin penetration in the tissue microenvironment after TSL intravascular release (a, c, e, g). Red channel with AngioSenseTM (blood-pool labeling) showed that the acquisition was performed in a steady FOV and that the vessels were functional (b, d, f, h, i). The green channel shows DOX fluorescence signal enhancement after the bolus injection of TSL.
The fluorescence signal enhancement was present in the extravascular space, as well as in the cell nuclei of the tumor interstitium, which reflected intracellular uptake of released DOX. Interestingly, the onset of DOX uptake in the cells started from one side and progressively spread to involve cells throughout the FOV (Figures
Real-time monitoring of released-DOX penetration in the tumor interstitium shows cell nuclei that display increasing fluorescence signal (Figures
Uptake rates 1/
The spatiotemporal distribution of the released-DOX native fluorescence directly reflected released-DOX penetration (Figures
Assessment of released-DOX penetration derived from the implemented fluid dynamics model. (a–e) Fluorescence images measured at 3 min 10 s (a), 3 min 40 s (b), 4 min 10 s (c), 5 min 20 s (d), and 13 min 10 s (e). (f–j) Corresponding released-DOX penetration shown by the displacement field. For an easier visualization, only displacement vectors associated to voxels with sufficient fluorescence signal (i.e., greater than 5% of the maximum fluorescence signal intensity) are displayed. The principal component analysis allowed determining the main direction of released-DOX penetration, with 94% of relative displacement along the eigenvector 1 (k). The temporal profile of relative displacements along the eigenvector 1 equilibrated 7 minutes after the fluorescence onset (l).
The principal component analysis allowed determining the main direction of released-DOX penetration, with 94% of relative displacement along the principal axis 1 (Figure
Released-DOX penetration was modeled using the evolution model of equation (
Modeling of released-DOX penetration in the tumor microenvironment. The implemented evolution model allowed calculating a drug diffusion of
After monitoring dynamically DOX penetration after intravascular release at a single location, manual exploration of the tumor surface by the FCFM probe showed heterogeneity of DOX distribution in the tumor (Table
Evaluation of drug penetration in the tumor microenvironment is key to optimize therapeutic strategies of hyperthermia-triggered drug delivery in solid tumors. In this feasibility study, a setup was devised to monitor DOX penetration in the tumor interstitium in real time after intravascular release of DOX from TSL. Here, we used FCFM as an alternative of commonly used dorsal skinfold window chambers to assess drug penetration in a mechanically unrestricted tumor microenvironment.
First, no apparent microstructural changes could be observed in tumor tissue exposed to a 43°C mild hyperthermia, thus confirming the noninvasive nature of the hyperthermia procedure. Real-time imaging using FCFM was performed at the tumor surface with a skin incision and allowed imaging separately the vascular compartment and released DOX after intravascular release from the TSL.
In the TSL group, the real-time monitoring of DOX was successful in 2 out of 5 cases in the TSL group only, but DOX was found in every tumor when exploring the tumor rim at the end of the imaging session. In the free DOX group, no DOX could be detected, neither during the real-time monitoring nor during the tumor exploration at the end of the imaging session, suggesting that the DOX concentration was too low to be detected in the extravascular space. This could be explained by a low concentration gradient between the tumor vasculature and the extravascular space, thus limiting free DOX penetration, hence the use of TSLs.
Released-DOX penetration kinetics could be assessed upon imaging of cell-uptake kinetics of released DOX in the tumor interstitium and DOX diffusion coefficient, both in the same dataset. Here, 241 cell nuclei could be detected using the image processing pipeline developed previously [
Using the transport model of equation (
It is important to underline that any spatiotemporal intensity variations occurring between timepoints
The diffusion coefficient reflects the magnitude of driving force generated by the concentration gradient built up between the vascular compartment and the tumor interstitium. The model of released-DOX penetration yielded a diffusion estimate around 2500
Another source of bias in the evaluation of the diffusion coefficients may here arise from the 2D observation of a 3D diffusion process; this 2D observation is due to the design of the FCFM. Hence, only a 2D implementation of equation (
After monitoring dynamically DOX penetration at a single location, the manual exploration of the tumor surface and its depth by the FCFM probe showed a clear spatial heterogeneity in DOX native fluorescence, despite the presence of tortuous vessels and the homogeneous heating ensured by the water bath.
Despite the detection of functional vessels with AngioSense, no real-time DOX penetration could be observed by FCFM in 3 out of the 5 animals in the TSL group whereas histopathology showed high DOX concentration in the tumor area in all animals (
Our FCFM and data processing approach could also serve for the study of extravasation of drug nanocarriers, e.g., its enhancement by hyperthermia [
FCFM can provide real-time visualization of DOX
The data used to support the findings of this study are available from the corresponding author upon request.
The authors declare that they have no conflict of interest to disclose.
This study was funded by the ERC project 268906 “Sound Pharma” (Prof C. Moonen). The authors are grateful to Burcin Ozbakir (Pharmaceutical Sciences Institute, University of Utrecht, Utrecht, the Netherlands) and Dr. Noboru Sasaki (Image Sciences Institute, University Medical Center Utrecht, Utrecht, the Netherlands) for their technical support. The authors also thank Prof. Gert Storm for fruitful discussions (Pharmaceutical Sciences Institute, University of Utrecht, Utrecht, the Netherlands).
Table S1: summary of the results rat-by-rat. DOX concentrations in urine and blood correspond to the results of single quantitative measurements. Figure S1: representative profiles of rat rectal temperature recorded during real-time monitoring. Figure S2: micrographs showing the impact of mild hyperthermia on the tumor microstructure in the proliferative area (b, c, i, j), the transition area (d, e, k, l), and the necrotic area (f, g, m, n) of the rat 4. No morphological damage was observed at 43°C (h–n) compared to 37 C (a–g) in any of the 3 areas. Figure S3: real-time monitoring of doxorubicin penetration in the tissue microenvironment after free DOX intravenous injection (a, c, e, g). Red channel with AngioSenseTM (blood-pool labeling) showed that the acquisition was performed in a steady FOV (b, d, f, h). No DOX fluorescence signal enhancement could be observed in the green channel after free DOX intravenous injection. Figure S4: micrographs collected after dynamic monitoring. In the interstitium, nuclei that take up doxorubicin (b) are predominantly located around the vessels (a). Figure S5: micrographs acquired after dynamic monitoring. Despite the presence of tortuous vessels characteristic of tumor tissue (a), a nonnegligible amount of areas in the tumor tissue did not show any doxorubicin signal (b). Figure S6: representative fluorescence micrographs (mosaicking) of a tumor tissue exposed to TSL. DOX distribution is heterogeneous. Objective: 10x.