A high frequency ultrasound aided study of kinetics of drug delivery in tumor models

The in vivo performance of a Fluorescence Molecular Tomography system as a function of pathophysiological parameters that determine the penetration of nonbinding fluorescent nanoparticle was examined through imaging of a series of three tumor models. The pathophysiological parameters examined were, vessel density, interstitial fluid pressure (IFP), and collagen content. Drug delivery and IFP were measured in vivo via fluorescence spectroscopy and a fiber-optic coupled pressure probe. Vessel density and collagen content were determined ex vivo through histochemical analysis. The kinetics of the 40 nm,10000 KDa, fluorescent particles, which were injected into the tail vein of the mice, was monitored by sequential excitation of the tissue on and off the tumor site through employment of sixteen source detector pairs interspersed linearly in reflectance geometry. Each optical fluorescence data set was collected at discrete time intervals in order to monitor drug uptake for a period of 45 minutes. The kinetics of the drug delivery and the average nanoparticle uptake were correlated with the vessel density, interstitial pressure and collagen content. The results of the correlations were verified to be consistent with the published relationship between the three pathophysiological parameters and nanoparticle drug delivery. 1.0 INTRODUCTION An ultrasound-coupled handheld-probe-based optical fluorescence molecular tomography (FMT) system has been in development for the purpose of quantifying the production of PPIX in (BCC) in vivo. The design couples fiberbased spectral sampling of PPIX fluorescence emission with a high frequency ultrasound imaging system, allowing regionally localized fluorescence intensities to be quantified[1][2]. The optical data are obtained by sequential excitation of the tissue with a 633nm laser, at four source locations and five parallel interspersed detection locations. This method of acquisition permits fluorescence detection for both superficial and deep locations in ultrasound field. The optical boundary data, tissue layers segmented from an ultrasound image and diffusion theory are used to estimate the fluorescence in tissue layers. The modularity of the system permits the employment of different filters; which would enable the measurement of fluorescence signals from a variety of fluorescence agents with near infrared emission peaks, including nonbinding fluorescence nanoparticles, used in this study to simulate cancer therapeutic agents. In general, the effectiveness of cancer therapy in solid tumors depends on adequate delivery of the therapeutic agents to tumor cells. Researchers have demonstrated that drug delivery to solid tumors involves multiple processes, including transport via blood vessels, transvascular transport and transport to through interstitial space [3]. In this section we briefly discuss three, tumor-characterizing, pathophysiology that dictate these processes and how researchers have related these parameters to delivery of therapeutics in solid tumors. One of the most important preliminary determinants in delivery of therapeutic agents to solid tumors is blood supply [4][5][6].Tumor cells must recruit and reroute vasculature in order to meet increasing demands of a growing tumor. The resulting vessels differ from the vasculature in normal tissues both functionally and morphologically; tumor blood vessels are generally more heterogeneous in spatial distribution, larger in size, and more permeable due to discontinuities of the endothelium [7][8]. This feature permits drug transport across tumor micro-vascular wall by extravasation via diffusion and/or convection through the discontinuous endothelial junctions [3] . This difference in vascular permeability between tumor and normal tissues partly explains the phenomena that high molecular weight Medical Imaging 2012: Ultrasonic Imaging, Tomography, and Therapy, edited by Johan G. Bosch, Marvin M. Doyley, Proc. of SPIE Vol. 8320, 83200Y · © 2012 SPIE CCC code: 1605-7422/12/$18 · doi: 10.1117/12.911954 Proc. of SPIE Vol. 8320 83200Y-1 Downloaded From: http://spiedigitallibrary.org/ on 07/23/2013 Terms of Use: http://spiedl.org/terms drugs are tumor selective [9] [10][11]. Another factor responsible for passive tumor-selectivity of high molecular weight therapeutics is the lack of lymphatic drainage in solid tumors. In other words, the enhanced permeability and retention (EPR) effect is governed by two mechanisms: lack of lymphatic systems and leaky vasculature [9] [12][13]. Another important determinant in delivery of therapeutic agents to solid tumor is interstitial fluid pressure (IFP) [14]. As discussed in the previous section tumor vessels are characteristically leaky, and yet a functional lymphatic outflow pathway is missing. As a result solid tumors have been reported to have a markedly elevated IFP compared with normal tissue [15][16][17][18][19]. This elevated IFP is hypothesized to be a major barrier in the transport of large macromolecules [20] [21].The reason behind this phenomenon is that transport of small molecules in interstitial space is mainly by diffusion, whereas transport of large molecules is mainly by convection [22]. The mechanism of diffusion relies on concentration gradient whereas convection depends on hydraulic conductivity and pressure difference [14]. Due to higher IFP in tumors compared to normal tissues, the net convection flow in interstitium is outward from the core of the tumor, preventing effective penetration of macromolecules inside the solid mass [23][3]. In the context of tumor micro-environment, another parameter which has been shown to affect the transport of macromolecules is the collagen level in the tumor [24]. Collagen is produced by fibroblasts in the extracellular matrix (ECM) and contributes to the tensile strength of the tissues [25]. High collagen level is responsible for the observed stiffness in tumor ECM [26]. Elevated levels of collagen are also an indicator of poor prognosis, metastasis and tumor reemergence [26] [27] [28]. Furthermore, the high collagen levels in the tumor ECM pose a significant barrier in the transport of a variety of therapeutics, ranging from small-molecular-weight drugs to nanoparticles [29] [30][24] [31]. Vascular density, elevated tumor IFP and elevated collagen level on the transport of fluorescence nanoparticles were quantified in three different tumor lines. Here we present the measured dependence of drug delivery on these three physiological parameters, and thus corroborate the performance of the FMT system in making bulk fluorescence measurements of drug delivery. 2.0 MATERIAL AND METHODS It has been shown that an increased levels of collagen in tumors, generally leads to an increase in palpation. Also a recent view suggests that IFP is actively controlled by tension on the collagen network exerted by the fibroblasts [14].Thus, selecting tumor lines with varying levels of palpation increases the chances of selecting tumor lines with varying degrees of IFP. Additionally, it has been demonstrated that high IFP causes the collapse of vessels in tumors [3]. Thus, selecting tumor lines with three different levels of stiffness, lends to selecting a set with three different levels of IFP and three different levels of vessel density. For this reason, each of the three tumor lines: human epithelial carcinoma (A431), Rat Gliosarcoma (9L) and human adenocarcinoma (AsPC-1), which have been verified to have different levels of stiffness to the touch, were chosen to be injected subcutaneously into the hind legs of three mice. The tumors were implanted with IACOOC approved protocol. To plant each tumor line,1 million cells and Matrigel® (#356231, BD Biosciences, San Jose, CA) (1:1) initially in complete media, were injected subcutaneously in 3 mice. The cells were allowed to grow for a period of two to three weeks, until they reached an average diameter of 4-10 mm. To simulate nanoparticle therapeutics, we injected the mice with 200 μl of 5% FluoSpheres® Carboxylate-Modified microsphere solution (Invitrogen) [32] . The size of the microspheres is on average 0.04 μm in diameter and 10000 KDa in molecular weight, with the excitation peak at 660 nm and emission peak at 680. To monitor the temporal dynamics of the microspheres in vivo, we employed the FMT system. Optical sources and detectors were placed in the reflectance geometry on the skin both covering healthy and tumor tissue. 2.1 FMT Instrumentation FMT, shown in Figure1a, has four source detectors interspersed linearly and fixed in a fiber probe holder. The source, a laser light exciting at 643 nm, passes through a 600-μm fiber to a short pass interference filter at 650 nm to prevent the excitation light from “leaking” into the detection line. The laser light, from the short pass filter, is routed through a 1by4 fiber switch (Piezosystemjena, Hopedale, MA) which serves to sequentially excite the target tissue at four locations. In detection, four USB-coupled spectrometers (Ocean Optics, Dunedin, Florida, model USB2000+) acquire the data in parallel, with wavelength resolution of .37 nm and spectral bandwidth of 540 to 1210 nm. A user controlled removable inline 650 nm interference long pass filter (Omega Optics, Brattleboro, VT) in combination with a stationary 655 nm absorbing long pass filters (Melles Griot, Rochester, NY CG-RG-655) passes the Proc. of SPIE Vol. 8320 83200Y-2 Downloaded From: http://spiedigitallibrary.org/ on 07/23/2013 Terms of Use: http://spiedl.org/terms fluorescence peak, while preventing the saturation of the spectrometers due to the excitation light. The removable interference filter enables the acquisition of transmission data, which is essential for calibration. Fig 1: Instrumentation: (a) Schematics of a FMT imaging setup. Light from a 643 nm laser diode is routed through a short pass filter to a one by four fiber switch to illuminate the target sequentially. Emitted light is subsequently routed through one or two (transmission data/fluorescence data) long pass filters to spectrometers. An Apple mac mini acquires and analyzes the data from spectrometers. (b) Ultrasound transducer is coupled to the optical probes via a fiber holder. (c) Schematics of Interstitial Fluid Pressure measurement setup. A 260 micron fiber optic coupled FabryPerot sensor is inserted half way through the solid tumor mass to measure the interstitial fluid pressure. The pressure data is then sent to an apple mac mini for analysis. Proc. of SPIE Vol. 8320 83200Y-3 Downloaded From: http://spiedigitallibrary.org/ on 07/23/2013 Terms of Use: http://spiedl.org/terms A spectral fluorescence data (SFD) set is composed of sixteen spectra, which are the result of sequential excitation of the tissue at four inline locations and parallel collection of the remitted light via four interspersed detectors. Acquired SFD is calibrated using the current standards in spectral analysis [33][34][35]. Dark data, spectral data collected in the absence of excitation light, is subtracted from SFD to eliminate the effects of background light on the data. The spectral auto-fluorescence data (SAFD), collected in the absence of injected fluorescence microspheres, captures the endogenous tissue spectrum. SAFD is then subtracted from its corresponding dark data. Since the pre and post injection position of the FMT probes on the tissue are the same, a direct subtraction of SAFD from SFD yields a decoupled fluorescence spectrum. Each decoupled SFD is subsequently normalized with respect to the corresponding transmission data, to correct for tissue heterogeneities. The area under the curve of the calibrated decoupled SFD is integrated to produce the fluorescence data. The fluorescence data due to tumor is averaged and normalized by the average of the fluorescence data due to healthy tissue to account for potential intramouse discrepancies in the number of injected microspheres. For each mice the temporal fluorescence data shown in Figure 2a, was acquired by collecting SFD sets starting three minutes after the nanoparticle injection at five minute intervals for forty five minutes. 2.3 IFP Traditionally, three methods have been used to measure local IFP: needle, “wick-in-needle” (WIN) and Micropipettes (MP), and a micropore chamber has been used to measure average IFP [20]. Each of these methods has its advantages and limitations, which are detailed in these publications [36] [20]. To measure interstitial fluid pressure, a FOP-M260 optical fiber coupled pressure sensor (Evolution, QC, Canada), with 0.2MHz sampling rate is used [37]. The system is composed of an optical Fabry-Perot (F-P) sensing interferometer which connects to a signal conditioner via an optical fiber. The signal conditioner sends a light signal to the F-P interferometer. A physical parameter, in this case pressure, alters the optical path difference (OPD) of the F-P interferometer. The OPD is encoded in the light wavelength escaping the interferometer and returning to the conditioner through the fiber optic. This signal is registered by a charge coupled device (CCD) in the signal conditioner [38]. In the absence of external pressure on the (F-P) probe end, the pressure reading was set to be 0 mmHg. The tumor was punctured with a 25 gauge needle, and the (F-P) probe was inserted halfway into the tumor. After insertion, the optical fiber connected to the probe was gently released so that the pressure on the chamber of the F-P interferometer is solely due to the IFP. In the absence of any external pressure, the pressure stabilized asymptotically to the IFP point. The raw data is filtered by a moving average filter of span five to eliminate pulse artifacts due to the breathing of mice. The asymptotic value of the filtered data is the measured IFP. 2.4 Vascular Density CD31 is a sensitive marker for tumor angiogenesis as it stains blood vessels, and demonstrates the presence of endothelial cells in histological tissue sections. Precooled slides (-200C), containing slices of the tumor tissue were fixed in 1:1 acetone: methanol for 15 minutes at -200C. After air drying for 30 minutes, the slides were rinsed twice for 5 minutes in PBS with gentle agitation. The region of interest on the tissue were delineated with a Pap Pen, and the slide was blocked with a solution of 5% (Fetal Bovine Serum) FBS, 1%BSA (Bovine Serum Albumine) in PBS (Phosphate Buffered Saline) for 30 minutes at room temperature. The slides were subsequently drained, and the anti CD31 (BD Bioscience) at 1:50 each diluted in 200 μl 1% BSA in PBS were applied to the slides. The slides were incubated overnight at 40C. Prior to microscopy, the slides are rinsed three times for five minutes; the coverslips are mouthed and sealed with nail polish. The acquired images are placed under a simple thresholding algorithm such that the saturated signals and background are eliminated, leaving the stained endothelial cells. 2.5 Collagen Content To quantify collagen content, tumors, formalin fixed immediately after removal, were paraffin fixed, thin sectioned and Masson’s Trichrome stained. Slices were viewed at 10x (BX51 microscope, Olympus) and imaged using a color RGB camera (Qcolor3, Olympus). All processing was performed using Matlab (Mathworks). Images were cropped to remove section fold-overs, seen as black lines, and tumor outer edges. Images were then converted to HSV (Hue-Saturation-Value) space and thresholds for hue [180:235 for blue], saturation [adapted to average saturation for each image], and value [adapted to average value for each image] were applied to isolate collagen content, seen as blue in Masson’s Trichrome stain. The area ratio of blue stain was then calculated to determine the Proc. of SPIE Vol. 8320 83200Y-4 Downloaded From: http://spiedigitallibrary.org/ on 07/23/2013 Terms of Use: http://spiedl.org/terms collagen content for each image. 6-20 images were analyzed for each tumor and resulting collagen ratios were averaged. 3.0 Results and Discussion Figure 3a shows the median drug delivered to the tumor lines at five minute time intervals, three minutes after injection. The average drug delivered to the tumor is shown in Figure 3b. As discussed before, therapeutics with large molecular weights have been shown to selectively accumulate in tumors. This well-known phenomenon was captured by the FMT system, as it recorded higher fluorescence signals for tumors than for healthy tissue. The observed drug dynamics for subcutaneous tumor lines A431 and AsPC-1 are in accordance with the effects expected from the analyzed pathophysiological parameters on nanoparticle delivery. The higher vessel density (Fig 4) of A431 relative to AsPC-1 expedites drug delivery, and the lower IFP (Fig 2) and percent collagen content (Fig 5) of A431 tumor lines pose a significantly weaker barrier to the penetration of the nanoparticles into the tumor. Fig 2: Interstitial Fluid Pressure. Summary of measured IFP data for the three tumor lines. Proc. of SPIE Vol. 8320 83200Y-5 Downloaded From: http://spiedigitallibrary.org/ on 07/23/2013 Terms of Use: http://spiedl.org/terms Fig 3: Nanoparticle Kinetics and Uptake (a) The ratio of Fluorescence in Tumor Tissue vs Normal tissue in median tumor for each tumor line is shown. (b) The temporal average of the florescence ratio for each tumor line is presented. a