Evaluation of tumoral enhancement by superparamagnetic iron oxide

Eur Radiol (2005) 15: 1369–1377 DOI 10.1007/s00330-004-2586-8

P-Y. Brillet F. Gazeau A. Luciani B. Bessoud C.-A. Cuénod N. Siauve J.-N. Pons J. Poupon O. Clément

Received: 24 May 2004 Revised: 26 October 2004 Accepted: 5 November 2004 Published online: 23 February 2005 # Springer-Verlag 2005

P.-Y. Brillet (*) . F. Gazeau . A. Luciani . B. Bessoud . C.-A. Cuénod . N. Siauve . J.-N. Pons . J. Poupon . O. Clément Department of Radiology, Hôpital Avicenne, 125 rue de Stalingrad, 93009 Bobigny, France e-mail: pierre-yves.brillet@avc. ap-hop-paris.fr Tel.: +33-1-48955851

EXPERIME NTAL

Evaluation of tumoral enhancement by superparamagnetic iron oxide particles: comparative studies with ferumoxtran and anionic iron oxide nanoparticles

Abstract This study was designed to compare tumor enhancement by superparamagnetic iron oxide particles, using anionic iron oxide nanoparticles (AP) and ferumoxtran. In vitro, relaxometry and media with increasing complexity were used to assess the changes in r2 relaxivity due to cellular internalization. In vivo, 26 mice with subcutaneously implanted tumors were imaged for 24 h after injection of particles to describe kinetics of enhancement using T1 spin echo, T2 spin echo, and T2 fast spin echo sequences. In vitro, the r2 relaxivity decreased over time (0–4 h) when AP were uptaken by cells. The loss of r2 relaxivity was less pronounced with long (Hahn Echo) than short (Carr–Purcell–Meiboom–Gill)

Introduction Thanks to their marked r2 relaxivity and high magnetic moment, ultrasmall superparamagnetic iron oxide particles (USPIO) [1] have been used as a negative contrast agent for MR lymphography [2, 3] and liver imaging [4, 5]. Recently, they have also shown promise for cellular tracking and tumor imaging [6, 7]. In vitro, a wide variety of cells have been labeled with USPIO, including monocytes [8], lymphocytes [9] and tumor cells [10, 11]. Cellular uptake correlates with endocytosis and thus with tumor malignancy [10–12]. Internalization of superparamagnetic iron oxide particles by cells induces decrease in r1 relaxivity [13]; however, changes in r2 relaxivity are still discussed [10, 13]. In vivo, Moore et al. [12] observed particle accumulation in a rodent gliosarcoma model, with preferential location in tumor cells. These authors suggested that tumor malignancy

echo time sequences. In vivo, our results with ferumoxtran showed an early T2 peak (1 h), suggesting intravascular particles and a second peak in T1 (12 h), suggesting intrainterstitial accumulation of particles. With AP, the late peak (24 h) suggested an intracellular accumulation of particles. In vitro, anionic iron oxide nanoparticles are suitable for cellular labeling due to a high cellular uptake. Conversely, in vivo, ferumoxtran is suitable for passive targeting of tumors due to a favorable biodistribution. Keywords Iron oxide nanoparticles . Relaxometry . Magnetic resonance imaging . Cell magnetic labeling . Tumor targeting

could be estimated by measuring tumoral enhancement after particle injection. However, no precise description of enhancement mechanisms has been published, notably concerning the respective roles of cellular uptake, and particle extravasation through the tumor interstitium. Extravasation through the interstitium has been implicated in the brightening observed with T1 sequences in tumoral lymph nodes [14] and necrotic myocardium [15], and is due to r1 relaxivity of particles. Interstitial extravasation has not yet been modeled, but may be of major interest for predicting the therapeutic efficacy of high-molecular-weight anticancer drugs. As it is thought that abnormal high pressure in the interstitium could be the main barrier to drug delivery [16], we postulate that the “targeted” drug would follow the biodistribution of the iron oxide. The aim of this study was to investigate the mechanisms involved in tumor enhancement by superparamagnetic iron

1370

oxide particles, using a particle with high in vitro cellular uptake (anionic iron oxide nanoparticles [18–21]) and an USPIO with high in vivo tumoral distribution (ferumoxtran [1–3, 14]). In vitro, we wanted to assess the changes in r2 relaxivity due to cellular internalization. In vivo, we investigated the kinetics of tumoral enhancement before and after intravenous injection of ferumoxtran (Sinerem) and anionic iron oxide nanoparticles.

Tumor model In vitro and in vivo experiments were performed on the human prostatic adenocarcinoma cell line PC3 provided by Dr. M.F. Poupon (Institut Marie-Curie, Paris, France). Cells were cultured in 225-cm2 cell culture flasks (Falcon, Germany) at 37°C with 5% CO2 in medium (MEMα; Gilbo BRL, Praisley, Scotland) supplemented with 10% fetal calf serum, 100 IU/ml penicillin (Gilbo BRL) and 100 μg/ml streptomycin (Gilbo BRL).

Materials and methods This study was conducted in three phases. We first compared cellular uptake of ferumoxtran and anionic iron oxide nanoparticles in vitro. We then developed a model of dynamic relaxometry in order to measure the effects of cellular uptake on r2 relaxivity. Finally, we studied tumor enhancement in vivo, in a rodent model, in order to record the kinetics of enhancement on both T1 and T2 MRI sequences and to correlate them with iron quantification. Particles Ferumoxtran (Sinerem; Guerbet, Aulnay-sous-Bois, France; Combidex; Advanced Magnetics, Cambridge, Mass., USA) is a dextran-coated particle undergoing clinical evaluation for lymphatic [1–3] and hepatic [4] imaging. It is composed of a monocrystalline magnetite and maghemite core of 4.3–6.0 nm in diameter, coated with low-molecular-weight dextran. The mean hydrodynamic diameter of the nanoparticles is 31 nm. Ferumoxtran was reconstituted by mixing lyophilized powder with 9.7 ml of 0.9% saline solution. The final solution contained 350 mM iron. The r1 and r2 relaxivities were respectively 23 and 53 s−1 mM−1 (20 MHz, 39°C) in 0.5% agar [22] and 20 and 90 s−1 mM−1.in water (1.5-T MRI) [23] The half-life after intravenous injection is 4 h in rodent blood [24, 25] and 24 h in human blood [1–3]. Anionic iron oxide nanoparticles [18] are a new class of product with high cellular uptake rates [19–21, 26]. They have no dextran coat and their surface is negatively charged. We used anionic iron oxide nanoparticles (Laboratoire des Liquides Ioniques et des Interfaces Chargées, CNRS UMR 7612, Université Pierre et Marie Curie, Paris, France) composed of a maghemite monocrystalline core 8 nm in diameter covered with meso-2,3-dimercaptosuccinic acid. The negative surface charge ensures the stability of the ferrofluid in aqueous solution, through electrostatic interactions. The mean hydrodynamic diameter of anionic iron oxide nanoparticles is 24 nm. A brown-reddish solution of 150 mM iron was mixed with saline solution to obtain plasmatic iso-osmolality. The r1 and r2* relaxivities measured by 1.5-T MRI were respectively 9.4 and 357 s−1 mM−1. Blood distribution and elimination half lives after intravenous injection in a rodent model were, respectively, 10 min and several hours.

Cellular uptake quantification Experiments were performed at 60% cell confluence. The culture medium was removed and 5 ml of RPMI 1640 medium containing L-glutamine (without sodium phosphate) (Gilbo BRL) and particles was added. Particle concentrations were chosen from preliminary experiments showing that uptake of ferumoxtran was much lower than uptake of anionic iron oxide nanoparticles so that the concentrations were 1.8, 3.6, 7.2 mM and 0.9, 1.8, 3.6 mM for ferumoxtran and anionic iron oxide nanoparticles, respectively [19]. After incubation for 80 min, cells were washed twice with PBS (Gilbo BRL), scraped free and centrifuged. Measurements were performed using magnetophoresis [20] and ferromagnetic resonance [20, 27]. Magnetophoresis is based on measurement of the speed of magnetically labeled cells in suspension through a field gradient. When magnetic force (due to the field gradient) and hydrodynamic resistance (due to cell motion within the culture medium) are equal, cell velocity is constant. The amount of iron can thus be deduced, being proportional to the speed of cells. For magnetophoresis experiments, cells were dispersed in RPMI medium at a density of 104/ml, and 0.4 ml of this suspension was placed in a 1-mm thick Hellma chamber. For each particle concentration, the velocity and diameter of 30 cells were calculated using video microscopy and NIH Image shareware. The iron content per cell was calculated for a known field gradient, medium viscosity, and magnetic moment of a particle, and was expressed as ng/million cells. Ferromagnetic resonance [27] consists of detecting the resonance of electron spins in the ferromagnetic lattice. The hyperfrequency absorption spectrum depends on the amount of constituting particles. The frequency of the oscillating field (B1) used for electron spin resonance was 9.25 GHz and the higher intensity of the variable static field (B0) was 2 T. For ferromagnetic resonance experiments, cells were suspended in 15 μl of PBS, and 2 μl of this suspension was placed in a disposable glass micropipette, which was itself placed in a quartz tube suitable for electron spin resonance experiments. Another 2-μl aliquot was used for cell counting. The iron content per cell was expressed as ng/million cells.

1371

Relaxometry Relaxation times (T2) were measured with a Bruker minispec spectrometer PC120 (Bruker, Wissembourg, France) at 20 MHz (0.47 T) and 37°C. T2 was obtained from a multiecho sequence (Carr–Purcell–Meiboom–Gill sequence: CPMG, 100 points, TR/TE: 1 s/3 ms) and from a monoecho sequence (Hahn Echo sequence: HE, 20 points, TR: 4 s, initial/final TE: 2/380 ms) with monoexponential fit. The bandwidth was set to broad for both sequences. As the CPMG sequence includes many refocusing 180° pulses, it is less sensitive to field inhomogeneities induced by particles than is the monopulse HE sequence [28]. Three different media were used for particles relaxivity measurement (Fig. 1), namely water (medium A), and beads for cell culture (Cytodex; Amersham Pharmacia Biotech, Piscataway, N.J., USA), without cell monolayers (medium B) and with cell monolayers (medium C). Beads were used to induce heterogeneous spatial distribution of the particles (compartmentalization). When cells were cultured on beads, cellular uptake induced particle clustering. The beads were supplied as a dry powder and were hydrated with PBS and sterilized before use. The mean diameter of the hydrated beads was 150 μm. Dilutions of 0.05, 0.1, 0.2 and 0.4 mM particles were tested with Medium A and Medium B which allowed measurements without saturation of the signal. For dynamic measurements in medium C, 2.5 ml of hydrated beads were added to 8×107 PC3 cells and 20 ml of culture medium, and were incubated for 4 days. Particles were added at a final concentration of 0.05, 0.1 and 0.4 mM (anionic iron oxide nanoparticles) and 0.05 mM (ferumoxtran). Measurements were made immediately after adding the particles, and 2 and 4 h later. The system was put in the incubator between each measurement so that the time kinetic were obtained. Values corresponded to the sum from added extracellular and subsequently internalized intracellular iron oxides.

Fig. 1 Relaxometry experiments with particles (stars) in three different medium: medium A contained water, medium B contained Cytodex beads for cell culture, medium C contained Cytodex beads and PC3 cells (the different elements are shown with different scales to improve clarity)

Relaxivities (r2, s−1mM−1) were calculated in each experiment and results of uptake measurements are presented as relaxation rates (1/T2, s−1). In vivo MRI experiments We used 33 male Swiss nude mice (Iffa Credo, Paris, France), in accordance with local ethical guidelines. General anesthesia was induced before all procedures, by intraperitoneal injection of ketamine hydrochloride (Imalgene, Merieux, Lyon, France) and xylazine (Rompun, Bayer, Germany). Mice were xenografted with 106 PC3 cells seeded subcutaneously, on both sides of the thorax. They were imaged at 5 weeks, when tumor diameter was about 5 mm. Particles were injected through the tail vein at a dose of 400 μmol Fe/kg (0.2 ml). This was based on preliminary experiments in six mice with increasing doses (0, 200, 1200 μmol Fe/kg), and was optimal to observe positive tumor enhancement on T1 sequences 24 h after injection. MR imaging was performed on a 1.5-T unit (Signa, General Electric, Milwaukee, Wisc., USA). After anesthesia, animals were placed prone over a dedicated coil, and an oil-filled tube was placed in the field of view to act as a reference phantom. One mouse died after anesthesia and was excluded from the study. Therefore our study population consisted in 26 mice which were imaged once (n=16), or twice (n=10): before injection (control, n=6), and at 1 h (n=3 ferumoxtran, n=3 AP), 3 h (n=3 ferumoxtran, n=3 AP), 6 h (n=4 ferumoxtran, n=3 AP), 12 h (n=2 ferumoxtran, n=3 AP) and 24 h (n=3 ferumoxtran, n=3 AP) after injection. T1-weighted SE images [300/14 ms (TR/TE)] were obtained with an 4×4-cm field of view, a 256×256 matrix, three excitations, and 3-cm thick contiguous slides. T2weighted SE images [2000/80 ms (TR/TE)] were obtained

1372

with an 8×6-cm field of view, a 256×192 matrix, one excitation, and 3-cm thick contiguous slides. T2-weighted FSE images [3000/80 ms (TR/TEeff)] were obtained with a 8×6-cm field of view, a 256×192 matrix, five excitations, and 3-cm thick contiguous slides. T2-weighted gradient echo sequences were acquired but could not be analyzed because of poor quality (low contrast-to-noise ratio). Tumoral iron measurement At the end of the imaging procedure, animals were killed by anesthetic overdose. Three control tumors and one tumor per injection time were surgically harvested, placed in plastic tubes and frozen. Tissue samples were weighed fresh and dried until constant weight. Dry samples were placed in closed Teflon reactors with 300 μL of pure nitric acid and heated at 80°C until complete dissolution so that measurements included both extracellular but also intracellular iron. Pure water (resistivity >18.2 MΩ) was then added to obtain a final volume of 3 ml. Total iron was determined using Inductively Coupled Plasma Spectrometry (ICP-OES) on a JY 24 spectrometer (Jobin Yvon, Longjumeau, France). Detection limit was 0.003 μmol/g wet and reproducibility was better than 5%. Certified Reference Material (Bovine liver, NIST no. 1,577b) was used as quality control: found value 3.38± 0.14 μmol/g (n=3), certified value 3.1–3.9 μmol/g.

Fig. 2 Cellular iron uptake with ferumoxtran (Sinerem), measured by ferromagnetic resonance (circle), and with anionic iron oxide nanoparticles (AP), measured by magnetophoresis (open triangle) and ferromagnetic resonance (FR, black triangle). Ferumoxtran showed low uptake (50.5 ng/106 cells), for the highest particle concentration (7.2 mM) in the medium and the most sensitive method (ferromagnetic resonance)

area under the curve was calculated using cubic spline interpolation with a commercially available software (SAS Institute, Cary, N.C., USA) and curves were compared using a Mann–Whitney statistical test adapted from Delong et al. [29].

MR images analysis

Results

Necrotic and cystic tumors were excluded (n=4). The signal intensity (SI) of 53 tumors was measured with NIH Image software after selection of the whole visible tumor. Tumor SI was divided by the SI of the oil phantom, yielding a relative intensity (RI) for each tumor. Finally, the mean RI of the tumors in uninjected control mice was determined (RImean). Enhancement (ENH) values were calculated for each contrast agent and time point using the following equation:

Cellular uptake

ENHð%Þ ¼

RI
RImean  100: RImean

Statistical analysis For statistical analysis, we took into account the dependency of the tumors in animals with more than one tumor, so that a mean result was calculated for each animal. When animals were imaged two times after injection, each measured was taken into account. In order to compare intensity of tumor ENH using SE and FSE T2-weighted sequences, a paired Student’s t-test was used. In order to compare tumor ENH induced by both particles, the approximate

Figure 2 shows cellular uptake of ferumoxtran and anionic iron oxide nanoparticles after 80 min of incubation, according to the particle concentration in the incubation medium. Ferumoxtran showed low uptake (50.5 ng/106 cells), for the highest particle concentration and the most

Table 1 Relaxivity (r2) measured at 0.47 T, using short (CPMG sequence; TE: 3 ms) and long time echo (HE sequence; TE: 2– 280 ms) in medium A (water), B (beads) and C (beads and cells at t0, t2 and t4) for ferumoxtran (Sinerem) particles and anionic iron oxide nanoparticles (AP) Relaxivity (r2) s−1 mM−1

Ferumoxtran (Sinerem)

AP

CPMG

HE

CPMG

Medium A Medium B Medium C

120 15

77 50

175 11 9 0.06 0.7

t0 t2 t4

The bandwidth was set to broad for both sequences

HE 131 45 69 31 37

1373

Fig. 3 a, b Relaxation rates (1/T2, s−1) measured in Medium C (beads and cells) with CPMG (short TE) and HE (long TE) sequences between 0 and 4 hours, with anionic iron oxide nanoparticles (AP 0.05, 0.1 and 0.4 mM, respectively; squares, diamonds and triangles) and with ferumoxtran (Sinerem 0.05 mM, circle) used as control

sensitive method (electron spin resonance). Anionic iron oxide nanoparticles showed avid uptake (8.3×103 to 2.77×104 ng/106 cells) with both methods (magnetophoresis and ferromagnetic resonance), for the different particle concentrations of the incubation medium.

Fig. 4 Intratumoral iron quantification by inductively coupled plasma spectrometry (ICP/OES) after injection of ferumoxtran (Sinerem, circle) and anionic iron oxide nanoparticles (AP, triangle) (1 point/ particle at 1, 3, 6, 12 and 24 h)

Fig. 5 a, b Tumor enhancement values (ENH) measured at MRI a SE T1 and b SE T2 results at 1, 3, 6, 12 and 24 h after injection of ferumoxtran (Sinerem, circle) and anionic iron oxide nanoparticles (AP, triangle). In b, the data point for ferumoxtran at 6 h was corrupted by a high standard deviation and was excluded as an outlier

1374

Fig. 6 a–f MR imaging of subcutaneously implanted tumors in mice after injection of particles. Upper row: a control T1-SE, b T1-SE 12 h after ferumoxtran, c T1-SE 12 h after AP. Lower row: d plain T1-SE, e T2-SE 12 h after ferumoxtran, f T2-SE 12 h after AP. Positive tumor enhancement (arrow) is noted on SE T1 sequence whereas negative tumor enhancement (arrow) on SE T2 sequence 12 h after injection of ferumoxtran (Sinerem), indicating extravasation of particles through interstitium (arrows). Lower tumor enhancement (arrow) on SE T1 and T2 sequence after injection of anionic iron oxide nanoparticles (AP), which indicates a lower tumoral distribution

Relaxometry In medium A (water), relaxivity (r2) values were higher with anionic iron oxide nanoparticles than with ferumoxtran (Table 1). With both particle types, r2 values were lower in medium B (beads) than in medium A, especially when measured with multiecho sequences (CPMG). In medium C (beads and cells), r2 values were similar to those in medium B immediately after adding the particles (T0), and decreased after 2 and 4 h of incubation, representing internalization of particles. This dynamic phenomenon was most obvious in terms of the relaxation rate (1/T2) (Fig. 3), and was observed with both sequences for the different particle concentrations of the incubation medium. The decline in r2 relaxation rates was higher with CPMG sequences and with the highest concentration of anionic iron oxide nanoparticles in medium C (38, 57 and 56% with CPMG; 22, 48 and 37% with HE, for concentrations of 0.05, 0.1 and 0.4 mM, respectively). Conversely, 1/T2 values did not change over time with a ferumoxtran concentration of 0.05 mM.

tumor tissue). As a percentage of the injected dose, uptake was maximum (0.48%) with ferumoxtran at 1 h. In vivo MRI experiments Results of tumor ENH observed at MR imaging are shown in Figs. 5 and 6. Tumor brightening was observed on T1weighted MR images with ferumoxtran and darkening on T2-weighted MR images with both particles. Tumor ENH was higher with ferumoxtran compared to anionic iron oxide nanoparticles during the 24 h following injection (P