Challenges in the development of magnetic particles for therapeutic applications

Abstract

Certain iron-based particle formulations have useful magnetic properties that, when combined with low toxicity and desirable pharmacokinetics, encourage their development for therapeutic applications. This mini-review begins with background information on magnetic particle use as MRI contrast agents and the influence of material size on pharmacokinetics and tissue penetration. Therapeutic investigations, including (1) the loading of bioactive materials, (2) the use of stationary, high-gradient (HG) magnetic fields to concentrate magnetic particles in tissues or to separate material bound to the particles from the body, and (3) the application of high power alternating magnetic fields (AMF) to generate heat in magnetic particles for hyperthermic therapeutic applications are then surveyed. Attention is directed mainly to cancer treatment, as selective distribution to tumors is well-suited to particulate approaches and has been a focus of most development efforts. While magnetic particles have been explored for several decades, their use in therapeutic products remains minimal; a discussion of future directions and potential ways to better leverage magnetic properties and to integrate their use into therapeutic regimens is discussed.

• Magnetic
• nanoparticles
• drug
• delivery
• hyperthermia
• targeting
• cancer

Introduction

The interaction between magnetic particles and externally applied magnetic fields encourages particle use in the detection, diagnosis, and treatment of various diseases. The following is a non-comprehensive survey of the scientific literature that from the author's viewpoint delineates key particle structural considerations and biological parameters that impact the utility of magnetic particles in therapeutic regimens.

Whether a material is magnetic or not, size plays a critical part in determining in vivo performance. Thus the effects of size reported for a variety of polymers and nanoparticles are introduced before delving specifically into magnetic particles. Particles are typically used with the intent to deliver a particular material to a cellular target in a particular tissue, but undesirably rapid kidney clearance and reticular endothelial system (RES) uptake mechanisms must be circumvented. Glomerular filtration will clear materials with a molecular weight of less than about 40 kDa, or about 10 nm [1]. For particles larger than this the main mechanism of particle removal from the bloodstream is via the RES. Unless RES uptake is specifically desired, any systemically administered particle must be designed to avoid rapid opsonization (blood protein binding) and subsequent macrophage phagocytosis. Coating particles with polyethylene glycol (PEG) is frequently employed for retarding opsonization, but eventually even these ‘stealth’ particles are taken up by macrophages in the liver, spleen, marrow and lymph nodes, for example as demonstrated for stealth liposomes [2].

Particles that remain at a high enough concentration in the blood stream for a sufficient length of time can drive a slow accumulation through the leaky vasculature found in a variety of lesions. Vascular leakiness is common to tumors and inflamed tissues characteristic of infections, arthritis, and artherosclerotic plaques [3], [4]. This ‘passive accumulation’ is perhaps the most cited attribute of macromolecular medicines and particulate drug delivery. In the case of tumors, in contrast to healthy tissues with tight junctions between endothelial cells, as tumors grow they develop a patchwork of defective endothelial cells which allow macromolecules and nanoparticles to extravasate [5]. Leakiness develops with angiogenesis that occurs surprisingly early in tumor growth, with Dewhirst and colleagues observing angiogenic processes in masses as small as a few hundred cancer cells [6].

In addition to dysfunctional blood vessels, tumors also have corrupted lymph vessels. Maeda documented what he termed the enhanced permeability and retention (EPR) effect, observing that macromolecules and other relatively large materials that leaked out of tumor vasculature remained for extended time in tumors compared to healthy tissue in part due to faulty lymph drainage [7], [8]. While many efforts have centered on leveraging the EPR effect, the advantage of large materials compared to small molecules in concentrating in the tumor's perivascular space becomes a disadvantage after extravasation. Due to the faulty lymph drainage, tumors have high interstitial pressures that retard permeation, and to reach cancer cells deep in tumor structures, material must permeate through a tumor's interstitial space. Additionally, tumors have heterogeneous, temporally dynamic structures with low porosity and dense extracellular matrices that provide substantial physical resistance to material permeation [9]. Tumor extravasation and subsequent permeation has been assayed in depth for a wide range of materials. Using carbohydrate materials, Dreher et al. found that dextran of 2 MDa with a corresponding hydrodynamic diameter of about 50 nm had an apparent tumor permeability of almost 100-fold less than 3.3 kDa dextran [10]. Dreher further observed that dextran polymers with molecular weights of 40 to 70 kDa, corresponding to a hydrodynamic size of 11 to 14 nm had maximum tumor concentration but remained close to the vessels from which they extravasated, while dextrans of molecular weight lower than 10 kDa penetrated deeply into tumors. In an earlier study examining tumor perfusion, Yuan et al. did not observe as large or straightforward an effect of tumor extravasation with material size as did Dreher. However, a variety of structures with different charges were surveyed by Yuan, a reminder that size is not the only important parameter influencing tumor penetration by macromolecules and particles [11]. Generally, hydrophilic, neutrally charged or better yet uncharged materials will stick less to cell surfaces and interstitial biomolecules. Yuan et al. also modeled the extracellular volume available to materials using multiple pore sizes and connectedness in the tumor extracellular space to explain the loss in permeability with increasing molecular weight [12].

Considering drug-conjugate polymer/particle structures, Lee et al. reported that small multivalent doxorubicin-PEG biodegradable ‘bowtie’ dendrimers with molecular weight in the 45 kDa range optimally treated tumors of a certain size. The authors speculated that the effectively treated tumors were large enough so that vascularization and concomitant vascular leakiness had developed, but not so large that the tumor could not be treated by the administered doxorubicin dose [13]. This result emphasized that the utility for particulate systems is in treating tumors that are vascularized. Importantly, the hydrolytically labile structure also allowed for a ‘cluster bomb’ effect utilized in various macromolecular pro-drug constructs, where particles are first localized to the tumor via the EPR effect, and then subsequently degrade to smaller sizes, for instance through enzymatic or pH driven hydrolytic mechanisms, to small entities that are better able to penetrate the tumor mass and access cancer cells in hypoxic regions [14]. Similarly for drug loaded liposomes, the liposome need not necessarily penetrate a tumor intact. Allen and coworkers found that relatively small liposomes of 80, 100 or 150 nm average diameter accumulated to a substantially greater level than 240 nm sized liposomes [15]. While liposomes have been extensively studied and developed [16], with doxorubicin loaded, PEGylated formulations one of the few systemically administered nanoparticles that have been commercialized, they are larger than desired for tumor penetration. However, as with pro-drug concepts, if the liposome payload can be released in the tumor vasculature or within the tumor space, the higher penetrability of small molecules compared to nanoparticles may improve therapeutic performance.

Magnetic iron oxide particles and their use as MRI contrast agents

For in vivo use, iron-based magnetic particles have an attractive combination of high magnetization, well-described metabolism and relatively low toxicity. The incorporation of iron oxide materials has been preferred over zero valent metallic iron in part because IO has better shelf stability and produces less oxidative stress induced toxicity in vivo. Also, IO particles under about 30 nm are superparamagnetic, with no magnetization in the absence of an externally applied magnetic field that can lead to aggregation. IO particles have been fabricated in sizes ranging from about 10 nm to 10 s of microns, with magnetic material, coating, and particle size all playing critical roles in determining particle magnetic properties and pharmacokinetics [17]. For in vivo use IO particles are best known as contrast enhancing agents for magnetic resonance imaging (MRI). MRI with iron oxide contrast agents combines magnetic fields with iron particles and provides important background information for the medicinal use of magnetic particles. There are many articles and books that cover the theory and practice of MRI [18] and contrast agent use in MRI [19]. Briefly, under the influence of strong magnetic fields, water protons in different tissues produce different magnetic signals, and this difference in signals makes a given tissue lighter or darker than surrounding tissue in an MR image. When IO particles are located in a given tissue they alter the magnetic field experienced by water protons and thus substantially change the signal coming from that tissue, providing better contrast. As with all nanoparticles, the tissue distribution of IO contrast particles is heavily influenced by size. Small superparamagnetic iron oxide (SPIO) agents that are about 50 to150 nm are typically rapidly taken up by Kupffer cells (RES in the liver), when administered parenterally. SPIOs are thus used to enhance contrast for liver pathologies. Ultra-small superparamagnetic iron oxide (USPIO) agents are 10 to 40 nm in size and can be formulated with hydrophilic coatings that slow opsonization and monocyte and macrophage uptake [20]. The extended residence time in the bloodstream of certain USPIO formulations makes them well-suited for vasculature imaging and also for imaging lesions where the particles can extravasate through leaky vasculature [19]. In the latter capacity, USPIOs can provide valuable information on cancer stage and potential for metastasis; increased neovascularization and vascular leakiness with tumor malignancy [21], [22]. Feruglose, one of the smallest iron oxide-based MRI contrast agents with a size of 10–20 nm and a blood half-life of 3 to 4 hours was useful in characterizing tumor permeability by Daldrup-Link et al. [23]. The particles were found to extravasate across the leaky vasculature of malignant tumors, but not across the intact endothelium of benign lesions. Similarly, in a rodent breast tumor model, feruglose enhanced MRI and provided accurate quantitative measures of tumor microvascular permeability that correlated with tumor grade [24]. Vascular leakiness has been quantified using kinetic analysis of dynamic MRI data [21], [25], [26].

In summary, the use of IO particles as MRI contrast agents has provided a good deal of information on the distribution and pharmacokinetics of sub-100 nm sized iron oxide particles coated with various materials. To explore the potential to identify biomolecular signatures using MRI, USPIOs have been functionalized with targeting agents and have demonstrated binding to biomolecular targets in vitro [27]. While targeting of USPIOs have been reported for tumor phenotyping in in vivo applications [28], [29], publications are relatively scarce. This is due in large part because with or without the display of targeting agents, particles passively accumulate and reside for extended times in diseased tissues as discussed above. The effects of a biomolecularly-specific targeting agent that are distinguishable from this passive, non-biomolecular specific tissue distribution are difficult to measure.

Magnetic particles used for therapeutic purposes

Therapeutic mechanisms of iron oxide-based nanoparticles can originate from the iron material in the particle, bioactive agents loaded into the particle coating or adsorbed on to the iron oxide surface, and from heat generated by the particles under the influence of alternating magnetic fields (AMFs). Also, the magnetic core may allow the particle to be concentrated in desired locations in the body by the application of a high strength magnetic field to those locations. Generally, relatively small particles less than about 100 nm are best for tissue penetration and AMF-driven heat generation applications, while larger particles can be more easily retained in static magnetic fields and are appropriate for magnetic capture. A variety of materials and methods have been used to coat ferrite materials to retard opsonization, improve biocompatibility, and enable or improve drug loading. For large micron-sized particles, cross-linked albumin [30] and PLGA [31] have been used, while for small iron particles, dextran or modified dextran coatings have been most often employed. PEG [32], pluronics [33], silanes [34] and lipids [35] are among other coating materials that have been used in attempts to provide formulations with good water dispersibility and stability in blood. PEG is particularly desirable due to its extensive track record for reducing opsonization and RES uptake for a broad spectrum of proteins and nanoparticles [36], [37]. A recent review of magnetic particle coatings is provided by Gupta [38].

Incorporation of therapeutic material into ferrite particles

Various bioactive agents may be incorporated into ferrite particles through physical or chemical means. In the case of treating iron deficiency anemia, the bioactive agent is provided by the iron in the iron oxide core. Pathologies that may be addressed by iron oxide nanoparticle products include radiation and chemotherapy induced anemia, hematological malignancies, congestive heart failure, post partum anemia, and inflammatory bowel disease. For treating iron deficiency anemia, intravenous (IV) injections have reduced side effects than orally administered iron supplements. Magnetic particles have an extensive record of use in vivo, and while the rate of metabolism of a particular type of particle depends greatly on the size and coating composition, in general it proceeds along a well described path where uptake into the RES leads eventually, for the iron in the core that is, to incorporation into the body's iron stores, mostly into hemoglobin in erythrocytes and ferritin in other cells [39]. Interestingly, animal ferritin is a supramolecular nanoparticle with an iron oxide mineral core that is similar in size to the smaller synthetic magnetic nanoparticles discussed below, but with low magnetization [40]. Ferritin has a ferrihydrite-like iron crystal core about 8 nm in diameter encapsulated by proteins, with the entire complex being about 12 nm in size [40], [41]. Still, magnetic iron particles pose toxicity concerns, especially in the particles containing metallic iron, because upon metabolism the iron can generate oxidative stress through harmful reactive oxygen species [42].

Many parenteral iron formulations have been used in people, with Venofer™ (American Reagent, Inc.) and Ferrlecit™ (Watson Pharmaceuticals) currently widely used in the United States. Venofer and Ferrlecit are both comprised of iron oxides, with Venofer an iron sucrose [43] and Ferrlecit an iron gluconate formulation. In both cases the iron material is in the form of an iron oxyhydroxide colloid of about 2 nm in size for Venofer and 3 nm for Ferrlecit [44]. Ferumoxytol, originally developed as an MRI contrast agent and now in development for treating anemia, has a bigger IO crystal and may be more easily administered and have less toxic side effects compared to currently commercialized formulations [39].

Beyond the therapeutic use of iron in the core, a variety of small molecule drugs have been loaded into iron oxide particles. For example, chemotherapeutics have been loaded into pluronic-oleate coated iron oxide crystals [33]. The ability to stabilize these constructs against rapid disintegration in the blood was viewed by the authors as a key to tumor accumulation when using IV injections. LHRH peptides have been incorporated into IO particles for tumor targeting, and lytic peptides have been incorporated into the same particles for membrane permeation and cell lysis [45], [46]. The anti-viral drug indinavir has been incorporated into magnetic particles that were taken up ex vivo by monocytes [47], [48]. The incorporation of small molecules and proteins into magnetic particles for capture by stationary, high-gradient magnetic fields localized to desired tissues is discussed below.

In addition to small molecule drugs, targeting agents may be attached to magnetic particles for therapeutic purposes. Potential targeting agents include vitamins like folate [49], peptides, and antibodies and antibody fragments [50]. Discussions of targeting agents used with nanoparticles have been widely published, for example Sapra and Allen have provided detailed reviews in terms of what has been accomplished for liposomes [51], [52], while a broader view of nanoparticle targeting is provided by Fahmy, et al. [53]. Biomolecules to which particular targeting agents may bind include receptors that are overexpressed on the surface of cells, vascular targets, and extracellular proteins such as fibrin. In addition to targeting agents, enzymes, including collagenase for cutting through the extracellular matrix (ECM) of a tumor [54], [55] and urokinase for fibrin clot busting [56], have also been attached to magnetic particle therapies.

Magnetic field capture of ferrite particles

Perhaps the most easily achieved magnetic manipulations of therapeutic value are ex vivo apheresis procedures where high strength, high gradient fields are readily applied to magnetic particles in biological fluids. In typical applications, magnetic particles are added to extracorporeal blood where they can bind to materials such as DNA, viruses [57], [58], stem cells [59], or cancer cells [60], [61] and then be removed from the blood using a magnetic field prior to reintroduction to the blood stream. Particles for these applications can be large to aid in magnetic purification and they need not evade the RES because they are removed from the blood prior to its return to the body. More ambitiously, extracorporeal toxin removal after particles are injected into the body as a means of faster and more complete detoxification has been proposed [62].

A more technically challenging application is to direct particles to tissues in vivo using high-gradient magnetic fields. As with size and biomolecular targeting, the objective of magnetic targeting is to better localize material to a lesion and avoid exposure of healthy tissues. With respect to cancer, while surgical resection is a first choice when tumor location is known, resection is not always possible or can be only incompletely accomplished, and magnetic targeting of drug-loaded particles to known tumor locations is potentially useful.

Among others, Driscoll, Widder, Senyei and colleagues [63] and Gupta and Hung [64] have performed experiments and presented calculations on field strengths and particle sizes necessary to retain particles in tissues and under the influence of blood flow. Force on a particle is a product of the externally applied magnetic field strength, magnetic gradient, particle volume (that is, the cube of the diameter), and particle magnetization. Thus the combination of large, high magnetization particles with high field gradients is necessary for retaining particles in magnetic fields. Strongly magnetic iron based particles can be either ferromagnetic or superparamagnetic. Ferromagnetic materials have strong remnant magnetization after a magnetizing field has been removed; metallic iron particles above a certain size fall into this category. Superparamagnetic materials on the other hand have strong magnetization under the application of an external magnetic field of sufficient strength but no net magnetization after the external field is removed; certain nanoparticulate iron oxide minerals, for example magnetite, can be superparamagnetic. In the absence of a magnetic field, superparamagnetic particles do not agglomerate through magnetic attraction; agglomeration leads to RES uptake as well as accumulation in healthy tissues like the lung and thus is generally not desired for systemically administered particles.

The magnetization for ferrite materials can range from about 100 emu/g for magnetite (Fe₃O₄) particles to over 200 emu/g for metallic iron, with values for a given material very sensitive to fabrication methods. A primary challenge, especially in deep tissues away from the magnet, in using magnetic fields to localize magnetic particles is in generating a strong enough magnetic force. Lubbe calculated that for permanent magnets, particles could only be concentrated in tissues that were within a few cm of an edge of typically available strong NdBFe permanent magnets [65].

Large particles for magnetic capture can be made from ‘top down’ methods, that is, starting from larger particles and physically reducing their size. In the case of particles that advanced into late stage human clinical trials by the company FeRx, iron-activated carbon particles, termed magnetic targeted carriers (MTC), were combined with a variety of chemotherapeutic agents, including doxorubicin [6]. As shown in Figure 1, these iron-carbon particles had irregular shape and broad size distribution in the 0.5 to 5 micron range and were formed by mixing iron power and activated carbon using a ball mill mixer and ethanol as a lubricant, then homogenized and dried. The large particles were not amenable to systemic administration because they generally interact too rapidly with blood proteins and are rapidly removed from circulation; instead they were infused into arteries, for instance into the hepatic artery to treat liver cancer, with a high gradient achieved by placing high strength permanent magnets in close proximity to the liver to retain the particles and localize the release of the chemotherapeutic payload. For micron-sized iron-carbon particles with large magnetic susceptibilities of 175 Am²/kg, 100 gauss fields were sufficient to retain particles in flow fields representative of the bloodstream [6]. Importantly, despite the large particle size, the magnetic forces were apparently strong enough to cause the particles to extravasate and embed in the tissue where the fields were applied [66]. Besides infusion into the hepatic artery, other organs, including the lung after infusion into pulmonary artery, were tested in animal models using several chemotherapeutic drugs, including doxorubicin, mitomycin c, gemcitibine and oxaliplatin physically adsorbed on to MTCs [67]. In human clinical trials, doxorubicin, the most fully investigated chemotherapeutic, was mixed with and physically adsorbed to the particles just before injection into patient [68]. This material was called MTC-Dox, and later branded Magnetarg™. While the results of early clinical trials were promising, a late stage clinical trial yielded disappointing results that led to the discontinuation of development of chemotherapeutic MTC product candidates.

Figure 1. Scanning electron micrograph of iron-carbon microparticles prepared by FeRx for magnetic targeting of doxorubicin and other chemotherapeutics. 300 × magnification. Reprinted with permission from Elsevier from Biomaterials 21 (2000) 1411–1420.

Senyei and coworkers similarly used roughly micron-sized particles comprised of magnetite, cross-linked albumin and doxorubicin [69], [70] as well as other chemotherapeutic and bioactive agents to magnetically direct chemotherapeutic activity to tumors with significant improvement in therapeutic index in animal models. However, these particles and concomitant use of magnetic targeting did not proceed into human clinical trials.

Starch-coated particles about 100 nm in diameter loaded with epirubicin, an anthracycline chemotherapeutic, were used by Lubbe and colleagues in multiple clinical trials [65], [71]. While good tolerability was indicated in a phase I study in man, improvements in treatment protocols and materials, particularly the use of bigger particles, were deemed necessary before continuing human trials.

Smaller still, PEG-functionalized magnetite particles reportedly 30 to 60 nm in size were decorated with urokinase and magnetically directed to a fibrin clot in vitro using a 250 Oe magnet [32], and this particle was further investigated in vivo [56]. Due to the small size of the particles the magnetic field has to be very high with a pole in close proximity to targeted tissue for effective magnetic localization. The authors noted that because magnetic particles are attracted to one another in the presence of the magnetic gradient, particle clumping in tissues may aid in magnetic field sequestering.

Magnetoliposomes about 200 nm in diameter with very high magnetite loading were formulated by Lesieur and colleagues [72] were magnetically targeted using a 0.3 T magnetic field and a reported gradient of 11 T/m applied externally to the tumors of subcutaneously inoculated tumor bearing mice [73]. Liposomes were observed to embed in the capillary walls of the tumor and to be retained. While the particles were not loaded with chemotherapeutic, the authors propose that chemotherapy and/or AMF-driven hyperthermia are subsequently possible after localization.

For magnetic capture the magnetic field has typically been applied for short duration, usually less than an hour. Particle retention in the desired tissue after magnetic field removal has been a challenge; after a field is turned off, particles and their payloads are able to leave the tissue in which they were concentrated. This may occur through multiple mechanisms including diffusion of the particle back into the bloodstream, clearance by the lymph system, and uptake into phagocytosing cells and migration of the cells out of the tissue. To keep the particle in the desired location, a stealth particle that is not taken up by macrophages, that is sufficiently physically located away from the bloodstream in the perivascular space but that also is not easily cleared through lymph drainage are all desirable. It is notable that the MTC particles developed by FeRx were comparatively large and irregularly shaped compared to other magnetic targeted materials in clinical trial use and fared comparatively well. These particles were too large to penetrate further into the tumor after the field was turned off, but they also were effectively immobilized in the tumor where the payload could be released and penetrate effectively as a small molecule. However, in summary, the efforts of several academic and industrial organizations which began in the 1970s have yet to produce commercial products. The challenge of producing stable, stealthy, highly magnetic particles with sufficient payloads and high localized field gradients such that an improvement in therapeutic index is demonstrated in people have thus far been too difficult to overcome to commercialize products.

Therapy through particle heat generation

Targeting with static, high gradient magnetic fields discussed above is intended to direct the bioactivity of magnetic particles and thus is a delivery mechanism. The generation of heat by particles is intended primarily to have a therapeutic effect via hyperthermia. Particles containing ferromagnetic material such as iron or superparamagnetic iron oxide, typically a maghemite (γ-Fe₂O₃) or magnetite (Fe₃O₄) crystal structure, can be heated using high strength alternating magnetic fields (AMFs) [74], [75]. AMF-driven magnetic particle heating to provide localized hyperthermic treatment of cancers was proposed in the 1950s by Gilchrest [74]. Hyperthermia via AMF-driven iron oxide particle heating has progressed into human phase II clinical trials in Europe for the treatment of glioblastoma and prostate cancer. In these trials the particles are injected intra-tumorally to treat localized disease [76], due in part to the insufficient heating that can be achieved given the particle's Specific Absorption Rate (SAR) and limited tumor accumulation via systemic delivery as discussed below. The particles employed are narrowly dispersed iron oxide crystals in the 10 to 15 nm size range that are coated with aminosilane and the AMF applicator can reach 18 kA/m.

Before continuing the discussion of heat generation by particles, a few words on hyperthermic treatments in general are warranted. Hyperthermia can be applied using a variety of methods, including whole body heating, high intensity and frequency ultrasound, microwave–radiowave EM energy and infusion of warm water into the peritoneal cavity. Hyperthermic dose, that is both the tissue temperature profile and the duration of treatment, can be varied substantially and with these variations produce distinct and different outcomes. Thus, generalizations regarding physiological and biological effects should be made with caution. On the high temperature end, thermal ablative methods that heat tissues to above 50°C for even short duration directly kill cancer cells [77]. At lower temperatures, mild temperature hyperthermia (MTH), that is a dose in the range of about 39° to 42°C for about one hour, produces more subtle effects, including induction of a heat shock response and immune system stimulation [78]. Perhaps most relevant to the consideration of particle penetration, MTH increases tumor blood flow and vessel permeability, as reported by a number of researchers, including Vaupel and coworkers [79–81]. MTH also generally increases tumor interstitial permeability and cancer cell uptake of a range of solutes, including oxygen and other low molecular weight solutes [80], [82], as well as antibody fragments [83] and larger materials such as liposomes [84], [85]. In studies in mice, tumor permeability persisted up to about six hours after the initial hyperthermia application [85]. Interestingly, in a study using liposomes, increased permeability was not seen in a heat-cool-heat cycle experiment with the second heat cycle applied within several hours of the initial heating, an effect called vascular thermotolerance [85]. Thermotolerance is not observed if the heat cycles are spread out by sufficient time, for instance one week. Dosing schedules of hyperthermia and chemotherapeutic administration are thus critical to reaping benefits from combining modalities.

While systemic hyperthermia may be beneficial in treating metastatic disease, heat is a rather non-specific mechanism for cancer cell killing, with potential drawbacks similar to chemotherapeutics, and thus could benefit similarly by targeting its effects to pathological lesions. If particles capable of generating heat can be substantially localized in a tumor, tumor-specific heating may be achieved. Magnetic field heating of tissue embedded particles is attractive in part because magnetic fields in useful frequency ranges can penetrate the body deeply and are not reflected, in contrast with the problems of E-fields at the boundaries of materials with different conductivities. Ferromagnetic resonance (FMR) of iron's unpaired electron is one mechanism by which magnetic particles may be heated. Generally higher frequencies in the range on 1 to 1000 MHz may be useful for FMR. Larger particles may be heated through hysteretic effects using alternating magnetic fields in the range of 50 kHz to single digit low MHz. The most fully developed technology appears to be AMF-driven heating of small, highly crystalline particles presumably via Neel or Brownian mechanisms. Neel relaxation, that is, the fluctuation of a crystal's magnetic moment over an anisotropic energy barrier, and Brownian relaxation to viscous losses due to particle reorientation in solution, are two mechanisms typically cited for heat generation of iron oxide crystals less than about 30 nm in diameter. Theoretically, at resonant conditions when the particle's characteristic relaxation time (e.g. through Neel relaxation) matches the applied AMF frequency, the rate of heat generation, often referred to as the specific adsorption rate (SAR) is proportional to (M_s² V) · (H² f), where M_s is the particle magnetization, V is the particle volume, and H is the amplitude and f the frequency of the AMF [86]. AMF amplitude and frequency are limited by the generation of eddy currents in the patient's body. Eddy currents increase with the square of the radius of the AMF-subjected volume. The volume of tissue to which a field is applied is thus an important consideration. An often cited, experimentally determined maximum safe AMF power level for application to humans is a product of the frequency times the magnetic field amplitude such that H · f < 4.85 × 10⁸ A/(m·s) for one hour AMF application using a loop diameter of 30 cm [87]. Additionally, frequencies above about 100 kHz and below about 500 KHz appear to be preferred for safety reasons. Using sub-MHz frequencies and amplitudes up to about 10 kA/m, synthetic IO particles with specific adsorption rates (SARs) greater than 500 W/g have been reported [86]. Hergt and colleagues report that the SAR of a cobalt particle was about 720 W/g at the size where transition between superparamagnetic and ferromagnetic behavior, while magnetite of about 35 nm in size from bacteria have reported SARs of about 1000 W/g, which is the largest value reported to the author's knowledge [86], [88]. In the case of AMF heating of synthetic IO crystals, particle SAR appears to increase for monocrystalline IO particles with a diameter up to about 20 nm, not including coating. For a stealth particle, 30 nm or larger appears to be a reasonable estimate of minimum total particle size expected for maximum heat generation.

Extraordinarily high power output is predicted for monodomain crystals with narrowly defined size when the size is matched to the AMF driving frequency. Narrow size dispersion may also be desirable for additional reasons, such as eliminating small particles that aren’t magnetic and/or eliminating larger particles that are unstable in solution and are poorly circulating. Commonly followed IO synthetic methods similar to that forwarded by Molday have often been employed [89], where iron oxide nanoparticles were made by combining ferrous chloride and ferric chloride in a basic solution from which Fe₃O₄ precipitates. These precipitates are commonly coated prior to agglomeration by carbohydrates, particularly dextran or modified dextran, that are either present in the reaction mixture or introduced afterwards to form aqueous colloidal dispersions. Typical iron oxide particles made in this manner have a size in the 10 to 20 nm range. While particles made via iron chloride reduction do not have the monodispersed size that appears to be advantageous for AMF-driven hyperthermia, magnetic separation can be used to select particles of a desired size from mixtures with a broad distribution. Alternatively, recently developed thermal decomposition procedures similar to those published by Hyeon et al. [91] produced iron oxide particles with standard deviation in size of typically less than about 5% of the particle diameter for particles of about 10 nm or less. Particle synthesis yielded iron oxide colloids with oleic acid coupled to the surface, presumably forming coordination bonds with iron atoms on the surface of the iron oxide core via the carboxylate moiety. These IO cores were solubilized in water either coating or replacing the oleate layer with a hydrophilic or amphiphilic polymer [33]. Among other decomposition methods, ferrite nanoparticle cores with narrow size dispersion have also be prepared by reacting metal acetylacetonates (acac) with long chain alcohols in the presence of oleic acid and oleyl amine using procedures as described by Sun et al. [92].

Returning to heat generation, the amount of heat needed to raise the temperature of tissue depends upon the volume, surface area, and vascularity of the tissue. Assuming a spherical tumor that has a certain concentration of particles, the amount of heat that can be generated scales with the number of particles, which equals the tumor volume multiplied by particle concentration. The heat can exit the tumors by convection and conduction through the tumor surface and through blood flow. In the case of the surface, the heat transport q = h · A · ΔT, where an average heat transport coefficient between the tumor and the surrounding tissue, A is the surface area, and ΔT is the temperature difference between the tumor and surrounding tissue. Heat removed through blood flowing through a tumor also scales with surface area. Thus heat generation scales with volume, or the cube of the diameter, while the heat dissipation scales with surface area, or with the square of diameter. Thus for a given concentration of particles, the temperature of a large tumor can be raised higher than a small tumor, and there is a minimum volume necessary to produce a temperature increase. At the lower end volume extreme, it appears that a single, isolated cell can not be heated even at high magnetic particle concentrations because the surface to volume ratio for a single cell is too great [93].

Andra, Hergt, Hilger and colleagues have measured and modeled AMF-driven magnetic particle heat generation in spherical tumors [94]; Figure 2 is adapted from the calculated particle heating data for a range of SARs [95], with the calculations modified to the lower particle concentrations in tumors expected for systemic dosing and to the lower tumor temperature increases needed for MTH. If about 2°C of temperature increase is produced in tissues by AMF application alone, then the particles need to supply enough heat to raise the temperature about 2°C for MTH. Particle dosing must be kept with in a safe range; previous USPIO MRI diagnostic agents were limited to 2.6 mg Fe/kg iron per injection. However, ferumoxytol, a newer formulation in late stage clinical trials for treating anemia, can be dosed at higher than 10 mg/kg [39]. If a dose of double this, or about 20 mg Fe/kg was acceptable for treating cancer, this would correspond to 400 µg of Fe for a 20 g mouse. Concentrations of 40 µg Fe per gram of tumor may be realized if 10% of the injected dose is delivered per gram tumor, which is at the high end of what is possible. As seen in Figure 2, in this particle concentration range, very high SARs of greater than 1 W/mg Fe are estimated to be needed to heat tumors of roughly 1 cc. Thus, while appearing to be eventually achievable with further gains in magnetic particle structure, a combination of lower toxicity and higher SAR than are currently available seem to be needed to treat tumors using systemically administered magnetic particles and AMF-driven heating.

Figure 2. Estimates of concentration of ferrite particles needed for a give tumor diameter to raise the temperature of the tumor 2°C. Both axes are log scale. Lines are shown for particles with specific loss power of 0.1, 0.3, 1, and 3 W/mg. The line indicating the maximum levels of ferrite particle tumor concentration anticipated to be achievable after systemic (e.g. IV) administration assumes an injection of approximately 20 mg/kg Fe. For a 20 g mouse this is 0.4 mg Fe injected and the line shows a tumor accumulation of about 10% of the injected dose per cc tumor. For IO particles with a given SAR, tumors of the size indicated by the intersection at the maximum concentration line or larger may be treatable. For example, IO particles concentrated at 0.04 mg Fe/cc tumor and with an SAR of 1 W/mg may heat tumors with diameter of roughly 20 mm or larger to a therapeutically useful temperature.

Success has been reported, however, using systemically administered magnetic particles and AMF-driven heating; in particular, a delay in tumor growth in a tumor bearing mouse study as reported by DeNardo et al. [96]. Particles were about 20 nm in size and made of dextran coated magnetite decorated with tumor targeting monoclonal antibodies; Very high field strengths of 80 kA/m at a frequency of 153 kHz were used in these studies, two orders of magnitude higher than the safety limit observed by Bezovich [87]. However, this high amplitude AMF was observed to be safe for mice when applied in pulses, for example with AMF field turned on for 90 seconds and then off for 90 seconds to allow for heat dissipation, with a total treatment time of up to 20 minutes. Denardo and colleagues speculated that isolated heating at a microenvironment level, perhaps at the cellular level, occurs to the extent that thermal ablation is achieved [97]. As discussed above, Hilger et al. among others argue that heating of such small volumes is not possible [98]. Perhaps some of the effects of AMF-particle interaction result in non-thermal effects on cell structure. For instance, Prasad and coworkers hypothesized that cell-killing ‘magnetocytolysis’ may occur through physical disruption of cellular membranes and perhaps other cellular structures [99], [100].

The above study notwithstanding, the SAR capabilities of currently available IO crystals, systemic administration of IO particles appear to at best approach the intra-tumoral concentrations needed for localized hyperthermia. Also problematic is that RES uptake is generally large, and the volume of the RES organs, especially the liver, is large. Due to lower surface to volume ratios larger volumes are heated more readily than smaller volumes as described above. Thus, avoiding undesirable heating of RES organs while still effectively heating small tumors may be a challenge. When the AMF application can be localized such that RES organs are avoided, undesirable organ heating may be avoided. Finally with regard to RES heating, RES accumulation suggests a potential for treating primary or metastatic disease of the RES organs, for instance liver cancer. However, the same problems with delivery of chemotherapeutic to RES macrophages is encountered–that is the macrophages are not the cells that we desire to treat.

Beyond ferrites containing only iron, thermal decomposition methods also facilitated the incorporation of divalent cations such as Zn, Mn, Ni, or Co into spinel crystal structures which has been shown to increase magnetization and SAR [101]. While gains in magnetization may be achieved, the complexity of synthesis and potential increase in toxicity, for instance when cobalt or nickel are employed, lessen the desirability of mixed cation ferrite particles. In another alternative to typical IO structures, iron core particles with iron oxide shell particles could be more effective in heating due to the higher magnetization of the metallic core. However core/shell structures are difficult to synthesize and to achieve the desired magnetic performance, large particles are often necessary. For instance Qiang and coworkers fabricated relatively large iron/iron oxide particles with the core and shells over a range of sizes in the 10 to 100 nm range, with particles larger than about 40 nm approaching the magnetization of metallic iron [102]. These particles are relatively large even without coating and thus will likely be difficult to solubilize and will have tumor penetration challenges.

Looking towards additional ways to leverage AMF-driven hyperthermia of magnetic particles, reports on the use of hyperthermia to improve chemotherapeutic delivery via phase change of a carrier material are instructive. Chilkoti and colleagues used a polymer exhibiting a Lower Critical Solution Temperature (LCST) to affect a hydrophilic to hydrophobic transition at 40°C; when a tumor is locally heated, the hydrophobic material adheres to the tumor tissue and is better retained [103], [104]. Particularly attractive is the combination of thermally labile liposomes with hyperthermia, with liposome disruption occurring below temperatures where healthy cells are harmed [105]. Heat-triggered release of liposomal doxorubicin has advanced into clinical trials for ovarian patients [106]. Combining concepts of thermally labile liposomes and magnetic liposomes, magnetoliposomes have been formed with thermally labile lipid bilayers [107]. Doxorubicin loaded into these particles was reportedly released in vitro with the application of alternating magnetic fields at 3.5 MHz and 1.5 mT field strength [108].

Improving particle penetration: Dynamic particle size and increasing tumor permeability

As has been emphasized throughout this review, the constraints that particle size place on the strength of magnetic interactions and extravascular tissue penetration confound the development and utility of therapeutic magnetic particles. Figure 3 summarizes nanoparticle size and structure effects on distribution and tumor penetration; particle size desirable for tumor accumulation is smaller than what is possible for long circulating magnetic particles with high SAR and far smaller than what is useful for magnetic localization.

Figure 3. Approximate particle sizes that appear to be best suited for (a) tumor penetration, (b) tumor accumulation, and (c) AMF-Neel relaxation-driven hyperthermia. Also shown are the size ranges for (d) making stealth IO particles for hyperthermia, and (e) particle size needed for magnetic field capture.

The in vivo modulation of material size is a means by which tumor penetration may be improved. This is part of the rationale for thermally labile liposomes, where chemotherapeutics are incorporated into liposomes and heat is applied to both increase tumor permeability and release the liposome payload [105]. AMF has been applied to labile magnetoliposomes as well to drive chemotherapeutic release [108]. However, if particle disintegration is to be driven by AMF-IO heat generation, then the concentration of particles in a tumor needed for the requisite temperature increase is likely to be too high for what can be achieved using large particles like liposomes.

In exploring ways to improve performance, it may be useful to view magnetic hyperthermia as a ‘pre-targeting’ concept. In drug or radioactive isotope pre-targeting, a binding agent is localized to a tumor prior to administration of a cell killing agent that binds to the pre-targeting agent [109]. In the magnetic particle case, particles are localized to the tumor prior to administration of the AMF. The more material that can be selectively loaded into the tumor prior to being energized by the AMF, the more heat can be generated. It may thus be beneficial to employ smaller magnetic particles than those with higher SAR so that the particles concentrate and deeply penetrate tumors.

In contrast to the size reduction concepts used in labile liposomes, if size could be increased to better adsorb AMF energy selectively within the tumor, this would allow a tumor specific platform to provide better absorption of magnetic energy for heating purposes. However, the application of particles prior to AMF application would not take advantage of increasing tumor porosity and particle localization through hyperthermia application immediately prior to particle administration, as shown by Kong et al. [84], [85]. Perhaps hyperthermia applied through non-AMF mechanisms immediately prior to particle parenteral dosing followed at some later time by AMF application would improve both particle localization and tumor specific heating, but such a regimen is perhaps too complicated to warrant investigation.

Instead of, or in addition to, manipulating particle size by employing polymer precursors or particle components to facilitate interstitial permeation, decreasing tumor permeability barriers is also conceptually appealing. For instance by displaying certain enzymes, particles may cut their way through the tumor ECM. Increased particle permeation through ECM material in in vitro experiments has been recently reported for nanoparticles displaying collagenase [54], [55].

Looking forward: Magnetic particle use in multimodal regimens

Multi-modal approaches are now the norm in treating cancer, with mono-modal approaches increasingly rare. Surgery is used to remove readily accessed and identifiable cancer lesions, followed by one or more of radiation, radiation sensitizers, chemotherapeutic ‘cocktails’ and targeted therapeutic agents like monoclonal antibodies. Specific combinations of these modalities produce superior outcomes for certain cancers and cancer stages; that is, certain modalities synergize with one another. For instance, platinum therapeutics have demonstrated synergy with the targeted therapeutics such as trastuzumab [110], [111]. Paraplatin in combination with paclitaxel has demonstrated promise as neoadjuvant and adjuvant in treating NSCLC. Platinum agents are also excellent radiation sensitizers [112]. Hyperthermia synergizes with a variety of materials, including arsenic trioxide [113], tirapazamine [114], cisplatin [115], and with radiation therapy [116]. Often, hyperthermia equipment is housed within radiation oncology facilities at major medical centers. Radiation produces oxygen radicals, which then produce double-strand breaks in cancer cell DNA which may lead to cancer cell destruction. Cancer cells can mitigate this damage by oxygen radicals by first detecting double strand breaks, then producing mediators that repair the breaks [117]. Multiple studies have shown that hyperthermia can substantially increase oxygenation levels in tumors, thereby facilitating radiation-induced oxygen radical formation. Additionally, hyperthermia reportedly both hinders the ability of cancer cells to sense DNA destruction and impedes DNA repair mediators. In a triple combination, hyperthermia, platin chemotherapeutics and radiation are particularly synergistic [118], [119]. The better localization of these modalities to the tumor environment may substantially improve the treatment of cancer.

Consideration of the enabling features of iron-based magnetic particles, attributes and capabilities of the various materials and methods that can be used to coat particles, and the loading of bioactive agents into the coating, has led to the fabrication of MagNaGel™ nanoparticles by the author and coworkers [120]. With size tunable in the 20 to 50 nm range and high hydrophilicity, MagNaGel nanoparticles avoid RES uptake and increase tumor penetration. Iron oxide cores were synthesized to optimize AMF-driven hyperthermia and allow MRI tracking. Examples of IO nanoparticle cores fabricated by Alnis are shown in Figure 4. SARs of 670 W/g of Fe (at 400 kHz and about 8 kA/m) have been achieved, approximately matching the highest reported values of synthesized IO in the literature; as discussed above, further improvements in SARs are desired. Platin chemotherapeutics were incorporated at greater than 10 wt%, and targeted agents were incorporated for targeting vasculature or target cancer associated growth factor receptors. These heat-generating, platin-loaded nanoparticles thus combine hyperthermia and platinum chemotherapeutics, and may improve radiation efficacy 121–123 and are being evaluated in in vivo studies.

Figure 4. Transmission electron micrographs of iron oxide particles made by Alnis BioSciences, Inc. The smaller particle shown on the left had an average size of 10 nm, stdev 0.8 nm demonstrating the tight size control currently achievable in IO particle fabrication, while the particles on the right had an average size of 23 nm, stdev 4 m and generated about 670 W/g when subject to a 400 KHz alternating magnetic field generated using a 2 turn, 2.5 cm diameter copper coil driven by Ameritherm Hotshot 5000 power supply supplying 200 amps (estimated magnetic field amplitude of 8 kA/m).

Conclusions

Interactions with localizing or energizing magnetic fields are integral to therapeutic applications of magnetic particles, and particle size plays a central role in determining the strength of these interactions. Iron or iron oxide cores with a minimum size of 100 nm for practical use with strong permanent magnetic fields appear to be needed for high gradient magnetic drug targeting, but this size is larger than desired for (tumor) tissue perfusion. The large particle sizes required may have limited the utility of high gradient magnetic field approaches to localizing drug delivery. For hyperthermia applications, smaller magnetic particles can be employed; coated magnetic particles that can be efficiently heated the 30 to 100 nm or larger. While such particles are larger than desirable for optimized tumor perfusion, the size still takes advantage of the EPR effect. According to theory and recently acquired data, AMF driven heating appears to be substantially improved by matching a particular frequency with a monocrystalline ferrite particle of a narrowly defined size. Further advances in particle design and construction that improve AMF-driven magnetic particle therapy to the extent that metastatic disease can be treated using systemically administered particles appears to be achievable.

In addition to size, particle stability and drug retention in vivo has been cited as a key for drug delivery using magnetic particles, and the lack of coating stability has hindered therapeutic product development [33]. Similarly, magnetic core coating stability has been cited as the key for improving hyperthermia performance [86]. So coating is forwarded as a key for both delivery and heating, and generally the formation of a stable, long circulating ferrite particle remains a technical challenge. Additionally, even for relatively simpler polymeric/supramolecular structures compared to magnetic particle constructs, the difficulties of building a stable structure to a desirable size, size tolerance, and disintegration profile are under-appreciated [13]. However, supramolecular assembly methods and materials are improving at the same time that biological knowledge is rapidly accruing, informing upon what materials can and should be combined to construct magnetic medicines. Looking to the future, the synergies gained by combining magnetic particles and AMF-driven hyperthermia with molecular targeting, drug delivery, and radiation may greatly improve treatments of certain cancers. Additionally, modality combinations that include magnetic particles may be beneficial in the treatment of inflammatory diseases and infections where the advantages of magnetic particle localization, particle heating and MRI tracking of disease location are desirable.

Acknowledgements

The author thanks Christopher Sunderland for synthesizing magnetic particles, obtaining the image shown in Figure 4 and for proofreading the manuscript. Also, the National Cancer Institute's support of Alnis BioSciences’ efforts under contract # HHSN261200411004C to develop magnetic particles for the detection, diagnosis and treatment of cancer is greatly appreciated.

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.