Superparamagnetic iron oxides as MPI tracers: A primer and review of early applications

Abstract

Magnetic particle imaging (MPI) has recently emerged as a non-invasive, whole body imaging technique that detects superparamagnetic iron oxide (SPIO) nanoparticles similar as those used in magnetic resonance imaging (MRI). Based on tracer "hot spot" detection instead of providing contrast on MRI scans, MPI has already proven to be truly quantitative. Without the presence of endogenous background signal, MPI can also be used in certain tissues where the endogenous MRI signal is too low to provide contrast. After an introduction to the history and simplified principles of MPI, this review focuses on early MPI applications including MPI cell tracking, multiplexed MPI, perfusion and tumor MPI, lung MPI, functional MPI, and MPI-guided hyperthermia. While it is too early to tell if MPI will become a mainstay imaging technique with the (theoretical) sensitivity that it promises, and if it can successfully compete with SPIO-based ¹H MRI and perfluorocarbon-based ¹⁹F MRI, it provides unprecedented opportunities for exploring new nanoparticle-based imaging applications.

Keywords: Cell tracking; Magnetic particle imaging; Magnetic resonance imaging; Molecular imaging; Nanoparticles; Superparamagnetic iron oxide; Tracer.

Figure 1:

The currently available two MPI systems. (A) Preclinical Bruker MPI system designed for 3D high temporal resolution real-time imaging applications with up to 46 volumes per second. With a bore diameter of 12 cm, the scanner is applicable for in vivo imaging from mice- to rabbit-size animals. The magnetic field generator is equipped with 7 individual electromagnets allowing for flexible sequence design by means of one scalable selection gradient filed of up to 2.5 T/m featuring a field free point (FFP), three orthogonal drive fields for particle excitation (~25 kHz and up to ±14 mT/±14 mT/±14 mT) and particle signal reception (bandwidth up to 1.25 MHz) as well as three orthogonal focus fields for slow field shifts (±18/±18/±42 mT). Imaging speed is maximized by using a FFP main magnet, giving an order of magnitude higher temporal resolution than imagers using FFL magnets. Together with the embedded ParaVision acquisition and processing platform, the MPI 25/20FF offers intuitive study planning, data acquisition and reconstruction as well as automatic system control. (B) Magnetic Insight Momentum MPI system designed for 2D high sensitivity imaging. With a bore size of 6 cm, the scanner is applicable for in vivo imaging of mice. The resolution of the system is driven by a selection field gradient strength of 6.2 T/m, featuring a transmit/receive subsystem (45 kHz drive, ±15 mT / ±15 mT) and slow shift fields of ±190 mT). Sensitivity is maximized by using a field free line (FFL) main magnet, giving an order of magnitude more signal than imagers using FFP magnets. System siting requires only power and cooling water, and does not require a shield room. The user experience offers one-click image acquisition and native DICOM support.

Figure 2:

First MPS measurements of stem cells labeled with commercial, clinically used SPIO formulations. Shown are the MPI signal amplitudes as a function of Fe content (A, B) and the corresponding number of cells (C, D). Data are shown for MSCs labeled with Resovist® (MR) and Feridex® (MF), and NSCs labeled with Resovist® (CR) and Feridex® (CF). Note the linearity of the MPS signal with both the iron content and number of cells except for the lowest concentration of the smaller NSCs (2,500 cells) that contain less iron. Reference samples (free, non-cell bound particles in gelatin) are included in A and B as open symbols, with no difference in signal from cell-internalized particles. Image reproduced, with permission, from Refs. [17, 19].

Figure 3:

First MPI/MRI cell tracking studies. MPI (A), MRI (B) and corresponding overlay MPI/MRI (C) of a mouse brain injected with 1×10⁵ (left hemisphere) or 5×10⁴ (right hemisphere) SPIO-labeled MSCs as a focal point transplantation in the striatum. Quantification of MPI signal measurements (D) show an excellent correlation between the measured MPI signal and number of implanted cells (~2:1 ratio). To this end, an intensity profile through the centers of both MPI signal peaks was extracted and fitted with two Gaussian curves. Five parameters were fitted, namely the amplitudes, positions, and one width parameter, which was assumed to be the same for both curves, as it is mainly determined by the extent of the calibration sample used for system calibration. Fitted amplitude values were 7.24×10⁻³ and 3.48 × 10⁻³ a.u. at 4.55 and 9.69 mm for 1×10⁵ and 5×10⁴ cells, respectively, with a full width (resolution) of 3.68 mm. Calculated AUC MPI SI ratios were 2.08:1. Image reproduced, with permission, from Ref. [19].

Figure 4:

MPI/CT cell tracking. MPI/CT imaging of intravenously injected hMSCs, with representative coronal, sagittal, and axial slices shown from full 3D MPI datasets. (A) MP imaging of hMSC tail vein injections less than 1-hour post-injection shows hMSC localization to lung only. (B) By 12 days, cells clear from the lung and migrate to the liver. Figure adapted from Ref. [21].

Figure 5:

MPI perfusion imaging. (A) SPIO tracer bolus passing through stroked mouse brain. Shown are slices of automatically fused 3D MPI/MRI data at several time points for the coronal, sagittal, and transverse planes. The ischemic hemisphere can be easily detected on MPI as a decreased perfusion area (red hash). Then occluded and non-occluded common/internal carotid arteries are indicated by a red asterisk and arrow, respectively. Blue arrow indicates the basil artery. MR dynamic susceptibility contrast (DSC) images are shown for comparison (bottom row). Reproduced with permission from Ref. [38]. (B) (Left). SPIO tracer bolus passing through a normal rat and a rat post traumatic brain injury (TBI). Note the large hemorrhagic brain area in the TBI animal, a result of vascular leakage. (Right) MP images obtained after three days post-injection. Blue and green circles indicate the area of the impact site and lymph nodes, respectively. Unlike the control, the TBI rat continues to have significant signal from the hemorrhage and with SPIO accumulation inside the lymph nodes. Reproduced with permission from Ref. [39].

Figure 6:

Brain fMPI. MPI measurement of ΔCBV modulated by a hypercapnic (5% CO₂) paradigm in a rat model. Ventilation properties controlled for alternating hypo/hypercapnia in 5 min periods. The MPI signal shows a 10% signal modulation with the expected CBV trend and CNR = 50. Representative blood-gas measurements were recorded for each condition, with respiratory rate/pCO₂/pO₂ values of 55 BPM/28 mm Hg/57 mm Hg, 32 BPM/49mm Hg/47 mm Hg, and 34 BPM/35 mmHg/79mm Hg for hypocapnia, hypercapnia, and baseline, respectively. Image adapted, with permission, from Ref. [50].

Figure 7:

MPI-guided tumor hyperthermia with sparing of normal liver tissue. During the standard MPI scan, negligible heating was observed in the mouse to the low frequency of 20 kHz. At a higher frequency of 354 kHz, all SPIO-containing tissues are heated up, including the healthy liver. When using MPI gradients, only the tumor is heated while the liver is spared. A dual tumor mouse model was then used to demonstrate arbitrary control of tumor heating. The FLL was first centered on the bottom tumor, heating the bottom tumor but not the top tumor. Without moving the animal, the FFL was then centered on the top tumor and the process was reversed. Image reproduced, with permission, from Ref. [55].

Figure 8:

Geographical overview of currently clinically used commercial SPIO formulations. Image courtesy of Dr. Eric Mayes (Endomagnetics, Ltd).