**Understanding MNPs Behaviour in Response to AMF in Biological Milieus and the Effects at the Cellular Level: Implications for a Rational Design That Drives Magnetic Hyperthermia Therapy toward Clinical Implementation**

**David Egea-Benavente 1,†, Jesús G. Ovejero 2,3,†, María del Puerto Morales <sup>2</sup> and Domingo F. Barber 1,\***


**Simple Summary:** Magnetic hyperthermia therapy is an alternative treatment for cancer that complements traditional therapies and that has shown great promise in recent years. In this review, we assess the current applications of this therapy in order to understand why its translation from the laboratory to the clinic has been less smooth than was anticipated, identifying the possible bottlenecks and proposing solutions to the problems encountered.

**Abstract:** Hyperthermia has emerged as a promising alternative to conventional cancer therapies and in fact, traditional hyperthermia is now commonly used in combination with chemotherapy or surgery during cancer treatment. Nevertheless, non-specific application of hyperthermia generates various undesirable side-effects, such that nano-magnetic hyperthermia has arisen a possible solution to this problem. This technique to induce hyperthermia is based on the intrinsic capacity of magnetic nanoparticles to accumulate in a given target area and to respond to alternating magnetic fields (AMFs) by releasing heat, based on different principles of physics. Unfortunately, the clinical implementation of nano-magnetic hyperthermia has not been fluid and few clinical trials have been carried out. In this review, we want to demonstrate the need for more systematic and basic research in this area, as many of the sub-cellular and molecular mechanisms associated with this approach remain unclear. As such, we shall consider here the biological effects that occur and why this theoretically well-designed nano-system fails in physiological conditions. Moreover, we will offer some guidelines that may help establish successful strategies through the rational design of magnetic nanoparticles for magnetic hyperthermia.

**Keywords:** hyperthermia; magnetic hyperthermia; magnetic nanoparticles; magnetic nanoparticleinduced biological effects; clinical trial; new therapies

#### **1. Introduction: From Cancer to Magnetic Hyperthermia Therapy via Nanomedicine** *1.1. Cancer*

Cancer is a multifactorial disease in which a variety of parameters influence its development, progression or outcome, such as the type of cancer, tissue localization, genetic predisposition, immune status of the patient, etc. For this reason, it is one of the most challenging diseases to treat and develop new and effective therapies, which in turn requires the cooperation of multidisciplinary teams. Many types of therapies have been approved to

**Citation:** Egea-Benavente, D.; Ovejero, J.G.; Morales, M.d.P.; Barber, D.F. Understanding MNPs Behaviour in Response to AMF in Biological Milieus and the Effects at the Cellular Level: Implications for a Rational Design That Drives Magnetic Hyperthermia Therapy toward Clinical Implementation. *Cancers* **2021**, *13*, 4583. https:// doi.org/10.3390/cancers13184583

Academic Editors: Moriaki Kusakabe and Akihiro Kuwahata

Received: 28 July 2021 Accepted: 7 September 2021 Published: 12 September 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

treat cancer and the specific therapy or a combination of these that patients receive depends on factors like the type or stage of development of the cancer. Traditionally, the most common treatments for cancer involve surgery, radiotherapy and chemotherapy. However, more recently, strategies like immunotherapy have been developed and implemented in combination with these established approaches due to their capacity to improve these treatments. Complementary therapies like stem cell transplant also help restore bloodforming stem cells in patients after particularly harsh treatments. In addition, biomolecular advances have helped us better understand the causes of certain types of cancer, guiding the use of more specific and precise treatments, for example, using biomarkers or genetic studies [1].

#### *1.2. Hyperthermia*

Hyperthermia (HT) is a cancer treatment strategy first shown to produce benefits in the 1940s when it was contemplated that rising the temperature of a tissue might combat fibrosis and cancer [2]. To treat cancer, HT involves exposing malignant tissues to supraphysiological temperatures [3]. Damage to tumour cells or their death are the main desirable effects of such heating but also, HT may improve tumour antigen presentation, the activation of dendritic and NK cells, and leukocyte trafficking through the endothelium [4]. These are phenomena that enhance the anti-tumour immune response, and that make cancer cells more sensitive to the effects of radiotherapy and chemotherapy. HT is generally considered to be defined as a rise in temperature of the tumour region to between 39 and 43 ◦C (known also as mild hyperthermia) [3,5], although an increase up to 45 ◦C may also be considered [6]. However, when the temperature rises above 45 ◦C the situation is usually referred to as thermal ablation, which may have dramatic side-effects due to the damage caused to normal tissue and the death of healthy cells. For this reason, careful temperature control is necessary during HT treatment. Usually, MHT is administered such a multivalent oncological strategy in combination with other anti-cancer approaches, specifically when HT has been demonstrated to produce an improved synergic effect [7].

HT therapies are mainly classified according to the area of the body treated, which is usually closely related with the method used to increase the temperature (energy source) [6,7].

#### 1.2.1. Whole-Body Hyperthermia

Whole-body HT is the systemic heating of the body in an attempt to obtain benefits treating widely disseminated metastatic cancer. The recommended upper limits are 42 ◦C maintained for 1 h, or 40 ◦C in combination with cytotoxic drugs. Technically, this method has important drawbacks, one of which is the need to sedate or submit the patient to general anaesthesia. Moreover, rising the body temperature from 37 ◦C to 42 ◦C is a lengthy process (90–180 min). In many cases the side effects of whole-body HT are unacceptable and much effort has been made to develop equipment that can resolve these disadvantages, e.g., IRATHERM-2000, currently in phase I/II clinical trials [8]. However, the success of this approach is limited by the poor balance between risk and benefit, and by the increasing interest in applying the fundamentals of HT in a safer and more specific manner [9].

#### 1.2.2. Regional Hyperthermia

Regional HT is the application of the HT to a whole organ, limb or region but not to a specific tumour area. The most commonly used strategy to raise the temperature in selected regions is perfusion HT. Perfusion HT consists of inducing a heated fluid flow, normally of the patient's own blood, through a specific area, and it is usually directed at tumours localized to a limb. Nevertheless, there are variants where stomach cancer or other tumours in the abdominal cavity are treated in this way, such as continuous hyperthermia peritoneal perfusion (CHPP). This approach is not technically difficult and it is safe to employ by heating the perfusion fluid up to 43 ◦C for 2 h. However, this technique alone does not produce outstanding results and only in synergism with cytotoxic drugs are the desired results produced. This combined therapy needs a precise adjustment of certain parameters, such as the flow rate (30–40 mL/min), the cells, pH and O2 of the blood perfused, or the amount of drug administered (which will be higher than in more specific therapies) in order to avoid undesirable toxicity [6,10]

#### 1.2.3. Local Hyperthermia

Local HT is based on the controlled heating of a specific tumour zone in an attempt to avoid side-effects in the healthy surrounding area. This approach is based on the application of electromagnetic waves or ultrasound through a physical stimulator, enabling local heat to be applied in an external or an invasive way. The external application operates through micro- or radio-waves directed at a superficial or slightly deeper solid tumour (only a few centimetres below the skin) by an external device (e.g., microwave antennas, radiofrequency electrodes, laser fibres, electromagnetic coils or ultrasound transducers). Alternatively, invasive local HT (or interstitial HT) requires a thin needle or probe to be inserted into the tumour and serve as in situ energy applier. In this case, invasive local HT can be used in a deep tumour but it is restricted to small tumours (less than 5 cm in diameter) located in an accessible organ or tissue, such as the head, neck, bladder or prostate. In general terms, the success of local HT is determined by the tissue characteristics and blood flow, factors on which the energy and heat distribution is strongly dependent. Indeed, the heat distribution is often not as homogenous as would be desired and to resolve this problem, segmented radio-frequency electrodes that allow a three-dimensional control of heating have been evaluated for clinical implementation [6,7,10].

#### 1.2.4. The Drawbacks of Conventional Hyperthermia

Without any doubt, HT is a very promising approach for cancer treatment and various clinical trials support its application for a wide range of different cancers: head, neck, melanoma, sarcoma, breast, glioblastoma multiforme (GBM), bladder, cervix, rectum, oesophagus, lung, mesothelioma and paediatric germ cell tumours [6]. However, since its first clinical application in the 1980s, its implementation has not been as widespread as might have been expected, probably reflecting several of the problems that traditional HT presents.

Firstly, HT requires specific equipment and technically, it is more complex than other more standard therapies like chemotherapy. Currently, there is considerable effort being directed at developing new, improved equipment. Secondly, the limited effectiveness of HT means it cannot compete with the standard protocols of cancer therapy. For an acceptable result, HT must be applied in combination with other therapies, mainly chemotherapy and radiotherapy, producing notable synergic effects. Consequently, HT is often considered as a sensitizing adjuvant therapy more than a cancer therapy by itself [11]. Traditionally, this reduced effectiveness can be explained by the uncontrolled dispersion of the heat, which is in turn caused by the lack of powerful devices to control and monitor the local temperature. In addition, this situation is exacerbated by the physical and physiological handicaps, such as the non-homogeneity of the tissues, the physical distance between the tumour cells to be treated and the heat source, or the thermal dissipation produced by the circulatory system. Thirdly, HT also produces side-effects and the rise in temperature may provoke non-desirable toxicity in healthy tissues or cells, even after local application. This is again due to the heat source not being exactly adjacent to the cancer cells, producing local side-effects. Finally, more studies into thermal biology and how HT affects individual cells molecularly are needed to better understand the process. Above all, it is necessary to assess the thermotolerance that has often been proposed [5,7,10] and to analyse the sensitivity of different types of cancer to temperature.

A general roadmap to address the aforementioned limitations involves improving the real-time temperature control of the tumour region and effectively localizing the induction of heat using a contactless stimulus. Regarding the former, technical developments have been quite successful, for example with Magnetic Resonance Thermometry (MRT). MRT is a MRI (magnetic resonance imaging) based technique that involves non-invasive

3D measurement of temperature distributions that could substitute the currently used invasive thermal probes. Since combining thermometry with MRI was first proposed four decades ago [12], MRT techniques have improved in accuracy and robustness for in vivo applications [13], now reaching pre-clinical stages of development [13,14]. Moreover, new approaches based on nanomedicine have been explored to obtain non-invasive heat sources that can be precisely targeted to the tumour cell in order to avoid side-effects and enhance the effectiveness of HT. Nevertheless, it would be desirable to have a single equipment where in vivo location and MHT could be performed simultaneously. However, limitations intrinsic to the technique (as different magnetic field requirements between MRI and MHT) do not allow a real-time guidance [15].

#### *1.3. Nanomedicine: A Trip through the Hyperthermia Based Nanotherapies to Treat Cancer*

Medicine is evolving towards more specific and personalised therapies, and HT must also move in this direction. Faced by this challenge, nanomedicine emerges as a very promising alternative to convert HT into a well-implanted and important cancer treatment. In recent decades, the application of nanotechnology and nanoparticles (NPs) to medicine and cancer has revolutionized techniques for diagnosis and treatment, now offering a broad range of alternatives. In terms of diagnosis, magnetic nanoparticles (MNPs) are contemplated as MRI contrast agents [16,17] or organic and inorganic NPs as nano-biosensors [18,19]. In addition, NPs also play an important role in cancer treatment, improving chemotherapy delivery [20–22], and they are being used in the development of innovative techniques such as gene [23,24] and HT therapies. Another interesting approach is the use of NP-loaded cells instead of individual NPs, a good example being the use of MNP-loaded immune cells for magnetic targeting in adoptive cell transfer therapies [25]. Both the possibilities for diagnosis and treatment make these nanomaterials, and specifically MNPs, very powerful candidates in the fight against cancer, know referred to as theranostic agents [16,26]. The success of NPs can be largely explained by the possibility of using them as more specific guided medicines and from the growing number of clinical trials, the use of NPs in cancer is becoming a reality [27–29].

First of all, NPs have the intrinsic ability to accumulate passively in tumours due to an Enhanced Permeability and Retention (EPR) effect, a concept coined in 1986 [30]. This EPR effect is based on the disruption and consequently, the loss of impermeability in the tumour vasculature, allowing the extravasation of proteins and macromolecules, and also of NPs into the interstitial space of tumours [31]. Moreover, the absence of functional lymphatic vessels contributes to the retention and non-clearance of the NPs [32]. The use of NPs (or macromolecular drugs, polymeric drugs or liposomes) is crucial to obtain an adequate EPR effect, since the use of these nanocarriers allows the desired drug to reach and accumulate in the tumour avoiding renal clearance caused by its small size (renal clearance threshold = 40 kDa) [30]. The success of the EPR effect is not exclusively dependent on the size of the nanoparticle or nanomedicine, but is much more conditioned by the tumour environment: size and concentrations of endothelial cells fenestrations, grade of fibrinolysis or thrombocytopenia in tumour hypoxic area, pericyte coverage of tumour microvessels, amount of collagen IV in basement membrane or density of extracellular matrix, and it, in turn, it is closely dependent of the type of tumour and organism. Nevertheless, an adequate diffusion extravasation and retention of NPs with sizes larger than 100 nm would be compromised [32]. Another way to direct NPs to cancer cells is through their active accumulation. This strategy involves the functionalization of the NP surface with antibodies or other specific ligands that recognize and promote NP uptake into target cells [33,34]. Even local NP injection into the tumour is a less aggressive and more specific minimally invasive option than conventional therapies not based on NPs [32].

Returning to HT, the possibility of a rationally targeting NPs to cancer cells offers two significant improvements over non-specific therapies. On the one hand, treatment effectiveness is notably enhanced due to the fact that each loaded NP can act as an individual heat source, only increasing the temperature in the areas where NPs accumulate and not

affecting the surrounding tissues. Consequently, the HT produced with NPs is associated with fewer side-effects. Moreover, research into NPs is still constantly refining their synthesis and preparation, which is becoming cleaner, faster and cheaper, as witnessed by their single-step synthesis by microwaves [35,36].

Consequently, nanomedicine is particularly relevant to the future of HT therapies, which are generally based on the accumulation of NPs at target sites and the application of an external stimulus that induces NP heating. There are four different ways to achieve nanomaterial-induced MH, being the nature of the stimulus and the type of nanomaterials used (with their intrinsic properties) crucial parameters to achieve hyperthermia phenomena (Figure 1; Table 1).

**Figure 1.** Therapies based on the accumulation of nanoparticles in the tumour and on the application of an external stimulus to induce nanoparticle heating: (**a**) Photothermal nano therapy (PTT), (**b**) Radiofrequency nano-hyperthermia (RFHT), (**c**) Ultrasound nano-hyperthermia (UHT) and (**d**) Magnetic nano-hyperthermia therapy (MHT).



#### 1.3.1. Photothermal Nano Therapy (PTT)

Photothermal nano therapy, usually shortened as photothermal therapy (PTT), is based on the capacity of agents to convert optical energy applied through a laser into heat. The laser beam can be tightly controlled, adjusting parameters such as the power density, duration or wavelength. Near-infrared (NIR) wavelengths (700–1400 nm) are normally used due to their good tissue penetration without producing undesirable damage or burns. Different nanomaterials can be used as photothermal agents but possibly, the best studied are gold NPs. When gold NPs are illuminated with NIR light, the energy produced by the excitation of their surface plasmon resonance (SPR) is transformed into heat and released to the local environment. The SPR effect is due to the oscillation of the free surface conduction electrons when excited by light, and they oscillate more with a resonance wavelength determined by the geometrical features of the gold nanomaterial [37,38]. For this reason, the size, surface and shape of gold-nanorods, and hence, their anisometry means they produce the best response to NIR excitation [39]. The PPT is a phenomenon shared by other nanomaterials, such as carbon nanomaterials (nanotubes, fullerene, graphene) [40,41] and CuS NPs [42] related to optical absorption. The effectiveness of PTT therapy is largely dependent on the penetration capacity of NIR, and it is more effective in melanoma and non-deep tumours. However, NIR-I (1000–1150 nm) penetrates well into tissues and organs like kidney, spleen and liver, whereas NIR-II (700–1000 nm) is the best option for muscles, and for stomach, heart and brain tissue in rats [43].

#### 1.3.2. Radiofrequency Nano-Hyperthermia Therapy (RFHT)

Radiofrequency nano-hyperthermia therapy (RFHT) is based on the capacity of different nanomaterials to absorb non-invasive radiofrequency electric fields (RF-EFs) and release heat in response [44]. The physical mechanisms explaining RFHT remain controversial and although Joule heating is considered the main mechanism, others will also contribute [45]. The longitudinal acoustic vibrational mode is another theoretical mechanism proposed to explain RFHT, for example involving the absorption of RF-EF energy by Fermi electrons [46]. Different nanomaterials release heat after exposure to RF-EFs, such as Pt or Si NPs and carbon nanotubes. However, the best studied are again Au NPs, for which high concentrations and small sizes have been demonstrated to be important parameters to enhance RFHT. The main advantage of this strategy is the greater penetration of RF-EF, which enables deep tumours to be treated. Indeed, RF therapies have been tested against bone marrow, liver, pancreas, colon or lung cancer [44].

#### 1.3.3. Ultrasound Nano-Hyperthermia Therapy (UHT)

Ultrasound nano-hyperthermia therapies (UHT) are based on the capacity of some nanomaterials to enhance the effects produced by exposure to ultrasound. Ultrasound therapy produces a non-specific response, such that extensive ultrasound exposure could provoke thermal damage to healthy tissues. However, high intensity focused ultrasound (HIFU, 0.1–1 kW/cm2) is a less harmful alternative [47]. The selection of nanomaterials used in UHT is crucial due to their importance to enhance the effects of ultrasound therapy where NPs are located. Ultrasound affects tissues in two ways: thermally and mechanically. In UHT the thermal interactions depend on the attenuation coefficient and thermal conductivity of the NPs. The first of these contributes to the absorption and scattering of the ultrasound waves so that large NPs provoke a major attenuation and therefore, greater thermal dissipation. In terms of conductivity, small metallic NPs are the best candidates to improve the thermal conductivity when cell are loaded with these NPs. On the other hand, NPs maximize the mechanical interactions, consequently the cavitation nucleation threshold decrease and cavitation phenomenon is induced, which causes mechanical cellmembrane damage and cell lysis. Theoretically, this therapy could be applied to all types of tumours without any restriction in terms of depth. However, in practice it has been little studied due to the lower specificity and the expensive equipment required, and it has

mainly been carried out with silica, gold and iron-oxide NPs against breast, melanoma or colon tumours [44].

#### 1.3.4. Magnetic Nano-Hyperthermia Therapy (MHT)

Magnetic nano hyperthermia therapy (MHT) is essentially based on the intrinsic ability of MNPs to respond to alternating magnetic fields (AMFs) by converting magnetic energy into heat [37,44]. This approach is a valuable alternative for some kind of cancers located close to vital organs and in particular, for those difficult to remove surgically like some brain tumours [48]. This therapeutic approach allows the main clinical limitation of conventional radiotherapy to be overcome, that is the lack of selectivity of ionizing radiation which damages healthy and tumour tissues alike [49], thereby limiting the treatment to certain tumours. In MHT, the tumour cells (in vitro)/tumour (in vivo) are first treated with functionalized MNPs that are specifically internalized by these cells and then, the cells/tumour are subjected to an AMF to increase the local temperature of the tumour cells and induce controlled apoptosis. The main disadvantage for this therapy is having pre-clinical implementation problems and it is necessary a well-understanding actuation mechanism in order to exploit its full potential. Concretely, these handicaps are discussed in the point 5 of this manuscript: nanoparticles aggregation and consequently loss of ability to releasing heat.

The purpose of this review has not been to carry out an exhaustive search of all the articles published in recent years in medical databases that include the term "magnetic hyperthermia therapy", but we have preferred to include those articles that we believe have contributed the most to understanding MNPs behavior in response to AMF in biological milieus, which is a critical step to drive magnetic hyperthermia therapy toward clinical implementation. The focus of this review is placed on the specific issues related to the physical basis of this therapy, the type of MNPs used and the tumours that are best suited to this therapy. The concepts underlying MHT will be addressed to understand what type of NPs are the best candidates to use. The biological effects of MHT at the cellular and molecular level will be explained and ultimately, the most promising and novel rationally improved strategies that currently produce the best results will be described. Finally, the current ongoing clinical trials will be reviewed and prospects for the clinical implementation of MHT therapies based on MNPs will be reviewed.

#### **2. Physical Concepts of Magnetic Hyperthermia (MH)**

The clinical application of magnetic hyperthermia is referred to MHT. This therapy is based on certain physical principles that we consider important to explain below. MNPs present unique magnetic properties that can be taken advantage of to achieve selective contactless heating mediated by AMF. The ferromagnetic materials used in the production of magnets are characterized by the remanent magnetization (MR) they present even in the absence of a magnetic field. This magnetization is a consequence of the alignment of atomic moments in a specific direction defined by the anisotropy of the material (easy magnetization axis). When ferromagnetic materials are reduced to the nanoscale, thermal fluctuations become more and more important until the thermal energy surpasses the anisotropy energy of the NPs, making the magnetic moment flip between two bistable positions of the easy axis and leading to what is called superparamagnetic behaviour. As a consequence of these spontaneous fluctuations, the MR of the MNPs disappears and the magnetization (M) presents a S-like reversible response to low frequency magnetic fields (H) like that indicated in Figure 2a. The time taken for the magnetic material to lose its magnetization is called the relaxation time [50].

**Figure 2.** Physical properties of magnetic nanoparticles (MNPs), and the magnetic response to low and high frequency magnetic fields. (**a**) At low frequencies the MNPs present a S-like magnetic response with a high susceptibility (superparamagnetic) and they transit towards a hysteretic magnetic response when exposed to high frequency AMFs. (**b**) The MNPs may lose spontaneously their magnetization by Néel and Brownian relaxation when the applied field is subtracted or rotate coherently in response to a rotatory magnetic field. Through these mechanisms the MNPs can convert the energy of magnetic fields into (**c**) heat, (**d**) mechanical vibration or (**e**) local torque in the cell environment depending on the features of the field applied.

MNPs recover a ferromagnetic response when the oscillation of the applied field is faster than the relaxation time of the magnetic moments (high frequencies) and the magnetization processes take place through dissipative loops. This dual response is of interest for their biomedical application as they become contactless nanoheaters. Thanks to the lack of the MR of superparamagnetic MNPs in the low frequency regime, they can be prepared as a colloidal suspension, avoiding aggregation. Consequently, they can be injected intravenously without any fear of them obstructing capillaries [51,52]. In addition, once they are situated in the target tissue they can be remotely activated as nanoheaters by applying a high frequency field, producing minimal effects in the surrounding biological tissues where only a weak magnetic response is produced. In contrast to other contactless mechanisms of activation, such as that required in photothermal or photodynamic therapies, AMFs can penetrate the body with minimal attenuation, ensuring homogeneous field conditions in the whole tumour without any shadow effect [53].

There are two dissipative mechanisms by which MNPs may lose their magnetization when the magnetic field is removed: Néel relaxation and Brownian relaxation (Figure 2b). The former relaxation is associated with the inversion of the magnetic moments between the two directions of the easy axis magnetization, and it depends on the magnetic features of the MNPs. The latter is produced by the physical rotation of the MNP within the liquid media. In both cases, relaxation leads to the misalignment of the easy magnetization axes and consequently, a cancelation the global magnetization of the system. Brownian rotation is a consequence of the random interaction with the surrounding media and therefore, it is controlled by the hydrodynamic size, the temperature and the viscosity of the medium [54]. In both cases, relaxation leads to the misalignment of the easy magnetization axes and consequently, a cancelation the global magnetization of the system. MNPs will adopt a faster relaxation mechanism depending on their intrinsic properties and those of the surrounding medium [55], although combinations of both mechanisms may exist. It has

been postulated that in certain AMF conditions MNPs may physically align their easy magnetization axis with the applied field before undergoing Néel relaxation [56].

It has generally been claimed that for translational MHT, MNPs design must focus on optimizing the mechanisms of Néel relaxation [57] since the natural MNPs-aggregation induced by contact with biological milieus (for instance, during lysosome encapsulation) and the high viscosity of these media blocks the Brownian effect [58]. However, the heat produced by MNPs with pure Néel relaxation requires tumour cells to have a high concentration of Fe in order to increase the temperature, which is difficult to achieve by intravenous injection and that has been resolved by intratumour injection [59].

An additional response of MNPs to an AMF is mechanical damage due to the magnetic torque generated in the presence of the field, which is in turn due to the misalignment between the field applied and the easy magnetization axis. This is a different concept to the Brownian relaxation indicated above. In this case, the magnetic moments of the different MNPs maintain their relative alignment but they are rotated collectively to reduce the angle between their magnetic moment and the field applied. Equation (1) shows how the magnetic torque (*T*) applied by a magnetic field (*B*) to a magnetic moment (*μ*) grows with the misalignment (*θ*) between them [60].

$$T = \mu \cdot B \cdot \sin(\theta) - 6\eta V\_H \frac{d\theta}{dt} \tag{1}$$

This equation also takes into account the resistance to rotation produced by the medium depending on its viscosity coefficient (*η*), the hydrodynamic volume of the MNP (VH) and the angular speed of rotation (*dθ*/*dt*). The order of magnitude of the mechanical torque in normal conditions for a single MNP is 10−<sup>21</sup> Nm [61]. However, the torque generated by the magnetic field is enhanced for an aggregate of MNPs [62], like those observed inside the cell's lysosomes.

The most relevant parameters (*η*, *VH*, AMF and frequency, etc.) have been revised in detail to determine the transition between heat dissipative mechanisms and magnetomechanical actuation. As a result, in viscous media the mechanical rotation induced by AMFs is mainly relevant for large MNPs (15–50 nm) and low frequency AMFs (<10 kHz) [63,64], although it can be exploited to exert effective mechanical torque on biological components [65]. This transition with AMF frequency becomes even clearer in the case of large anisometric NPs [63] and some empirical studies reported that AMFs induce a torque [66] or a physical movement of the MNPs [67] that may produce mechanical damage of the lysosome and cell membrane [68,69].

The use of rotatory fields is more convenient when inducing mechanical damage to cells. These fields maintain the MNPs magnetized throughout rotation and it is technologically simpler to generate homogenous fields of high intensity (>0.1 T) using permanent magnets. According to theoretical simulations, the torque induced by rotatory fields is 30-fold that created by AMFs [61]. Hence, many of the studies that seek to use MNPs to induce mechanical damage now focus on this kind of magnetic stimulus [66,70].

#### *2.1. Determining the Heating Power of MNPs*

One of the main limitations for the critical analysis of MHT measurements is how to determine the heating power of MNPs [71]. The Specific Absorption Rate (SAR) is an empirical parameter frequently used by radiological protection departments to regulate the amount of radiation absorbed by patients when exposed to radiofrequency fields and it is normalized to the mass of biological tissue irradiated in terms of W/g [72]. This parameter was taken by the MHT community to quantify the amount of magnetic energy transformed into heat by a suspension of MNPs. However, it is important to note that in the latter case, the temperature increase (ΔT/Δt) is generally normalized to the Fe concentration in the magnetic colloid (mFe) according to Equation (2), where CV is the specific heat of the colloid [73,74].

$$SAR(W/\mathcal{g}\_{Fe}) = \frac{\mathcal{C}\_V}{m\_{Fe}} \frac{\Delta T}{\Delta t} \tag{2}$$

It is also important to note that this is a system-dependent parameter that varies with the field intensity (H) and frequency (f) of the AMF. With the aim of standardizing such parameters and comparing the heating power of MNPs studied under different AMF conditions, an alternative parameter has been introduced, the Intrinsic Loss Power (ILP). This parameter divides the SAR by the frequency of the field and the square of the field intensity (Equation (3)).

$$ILP(W/\mathcal{g}\_{Fe}) = \frac{SAR}{f \, H^2} \tag{3}$$

This definition is based on the theoretical model proposed that considers a linear response of the magnetic moment to the AMF applied [75]. This model predicts a linear dependence of the SAR on the frequency and a quadratic dependence on H, although this is only valid for highly anisotropic MNPs and a small H, and thus, it cannot be assumed as a universal parameter. Besides, recent double-blind experiments showed that the specific features of the experimental set-ups and the thermal curves analysed may produce inconsistent SAR values between laboratories [71]. A promising solution to achieve a global and consistent parameter to determine heating power is to measure the high frequency magnetic loop of the MNPs [76]. As indicated previously, the amount of heat dissipated by a collection of MNPs is strictly related to the area of the high frequency hysteresis loops (A), see Figure 2a right). Therefore, the theoretical SAR can be derived from the product of this area to the number of magnetic cycles per second, i.e., the frequency of the AMF (Equation (4)).

$$SAR = A \cdot f \tag{4}$$

Hysteresis loops are intrinsic to magnetic systems and they provide information about the magnetic response to AMFs of different frequencies and intensities. Therefore, they provide information about the dissipative properties of the MNPs that are independent of the thermal diffusion of the medium and do not depend on the thermal loses of the calorimetric system. However, the AC magnetometers required to study these are still scare and generally homemade [77].

The ideal field and frequency AMF conditions for MHT are also still to be defined. The former limit was established for the maximum field-frequency product as H × <sup>f</sup> ≤ 4.85 × 108 A m−1s−<sup>1</sup> based on the feeling of discomfort in irradiated subjects [78], a subjective test of patient comfort. New AMF application systems can concentrate the radiation in a restricted volume, reducing the radiation dose received by the patient and enabling more flexible limits to be proposed, ranging between 1.8×109 Am−1s−<sup>1</sup> [79] to 8.3×109 Am−1s−<sup>1</sup> [80], and up to 18.7×109 Am<sup>−</sup>1s−<sup>1</sup> [81].

#### *2.2. Other Advantages of Using MNPs*

An additional functionality of MNPs that enhances their applicability in MHT is the possibility to concentrate them in a certain region using magnets. When MNPs are magnetized with a magnetic gradient they minimize their energy by shifting towards the region of the maximum field. This principle has been exploited to concentrate magnetic nanoagents to a target superficial tumour by locating a set of permanent magnets in the proximal skin area. Using the same principle, MNPs have been used to label circulating tumour cells so that they can be concentrated and detected for an early diagnosis of metastasis risk [82,83]. More recently, the magnetic guiding of immune cells loaded with MNPs has been proposed as an advanced solution to reduced vascular accessibility [84], and as a means to activate mechanosensitive membrane receptors that inhibit cancer proliferation [85].

MNPs also present interesting properties for clinical imaging techniques, such as MRI. The strong permanent field used to align the magnetic moment of water protons magnetizes MNPs, creating local regions of enhanced magnetic fields in the tissues where these MNPs lie. The local field created by MNPs modifies the relaxation time of the surrounding water molecules [86]. This changes the MRI contrast of the tissue loaded with MNPs and offers an interesting pathway for personalized therapy. In addition, Magnetic

Particle Imaging (MPI) has emerged as a promising technique to solve the incompatibility of a simultaneous MHT and MRI (mentioned in 1.2.4) and ideally achieve theragnostic NPs which are useful as heat generators for MHT, and at the same time, for diagnosis through real-time in vivo image during the therapy. MPI is an emergent image modality which works by detecting the nonlinear magnetization of the flipping MNP. MPI present several advantages: ideal penetration and signal-noise ratio, no view limitations, highly sensitive, linear and quantitative signal, high contrast, zero ionizing radiation, and safer and persistent. All this makes this technique an excellent non-invasive 3D tomographic imaging method to be combined with MHT for a real-time therapy image guided [15,87].

#### **3. MNPs for MHT**

Maximizing the heating power of MNPs is an interesting approach to minimize the dose required for effective HT therapy. In this section we will summarize the most important strategies to maximize the amount of heat dissipated by MNPs. Considering equation 4, it is easy to identify that the SAR can be maximized by increasing the frequency of the AMF, the area (A) of the high frequency hysteresis loops, or both parameters at the same time. The simplest strategy to increase A is to increase the magnetization of the sample. Taking magnetite (Fe3O4) as a reference for biocompatible magnetic materials, saturation magnetization can be increased by doping its crystalline structure with other transition metals. The magneto-crystalline structure of magnetite is an inverse spinel made of two sub-lattices of magnetic moments aligned in opposite directions that occupy octahedral and tetrahedral positions [88]. Due to the higher number of octahedral positions and the Fe2+/Fe3+ occupancies, the magnetic moment of these two sub-lattices is not compensated, which classifies magnetite as a ferrimagnetic material. Thus, in the presence of a magnetic field the octahedral and tetrahedral moments lie in parallel and antiparallel, respectively. As such, the global magnetization of magnetite can be enhanced by substituting Fe ions with other transition metals with a higher atomic moment, like Mn that occupies octahedral positions [89,90], or with transition metals like Zn that have no atomic moment and occupy tetrahedral positions. Both effects can be combined in ternary ZnxMn(1−x)Fe2O3 ferrites to maximize the saturation magnetization [91]. However, it must be borne in mind that the magnetic coupling between the lattices might be affected when high concentrations of dopants are used to reduce the magnetic order in the MNP, averting the enhanced magnetization. Besides, such complex formulations compromise the homogeneity of the stoichiometry in the sample.

An alternative strategy involves widening the magnetic cycle by increasing the coercivity field (HC) of the MNP. The Hc is related to the field required to cancel the remanent magnetization and thus, to the anisotropy of the magnetic system. MNPs with a high magnetic anisotropy (*K*) are harder to magnetize but they dissipate more energy as their magnetization is reversed. If the AMF is not sufficiently intense, most of the magnetic moments of the system remain fixed in their easy magnetization axis without dissipating any thermal energy. Thus, it is necessary to reach an AMF threshold to partially or completely overcome the anisotropy barrier of the MNP. The anisotropy field (*HK*) is an interesting parameter to define such a threshold in theoretical simulations (Equation (5)).

$$H\_K = 2K/M\_S \tag{5}$$

According to the numerical simulations based on the Stoner-Wolfarth theory, MNPs requires an AMF higher than approximately 0.4 *HK* to begin heat dissipation, which reaches it maximum value when the AMF approximates to the *HK* [92]. At higher AMFs, magnetic moments reach saturation and heat dissipation does not further increases. This theory also establishes a direct relationship between the HK and HC (HK = 0.48 HC) for a set of MNPs with randomly oriented easy magnetization axes, making this parameter an interesting link between theoretical simulations and empirical data [93]. In summary, increasing the magnetic anisotropy of the system can help increase the SAR as long as the intensity of the AMF applied is similar to the HK of the system.

The magnetic anisotropy constant (*K*) is a composite of two components: the magnetocrystaline anisotropy and the shape anisotropy. The former is associated with the crystalline structure of the MNPs and the coupling between their atomic magnetic moments. The magnetocrystalline anisotropy of magnetite can be enhanced by doping the structure with other transition metals, such as Co [94]. These cobalt ferrite MNPs have a K value that is 2 orders of magnitude higher than magnetite [95], creating MNPs with magnetic moments strongly fixed in the easy magnetization axes. Heat dissipation for highly anisotropic MNPs like cobalt ferrites is generally produced by Brownian mechanisms, which might be not ideal for in vivo applications.

The second contribution to anisotropy is defined by the geometry of the MNP. Magnetic materials tend to minimize their magnetic poles by locating their magnetization along their longest axis, establishing the geometrical easy magnetization axis. The geometric anisotropy can be enhanced by creating anisometric nanostructures like nanocubes, nanorods and nanodiscs [63,96–98]. These nanostructures are of general interest as they are good mechano-transducers that may apply large torques on biological components although they must be prepared with reduced dimensions to preserve their superparamagnetic response [99,100]. An alternative to superparamagnetism is to create nanomaterials with an exotic magnetic order known as magnetic-vortex that can be mechanically manipulated and also presents a lack of remanence [69].

As a rule of thumb, increasing the saturation magnetisation (MS) of the sample is a good approach for AMF inducers that operate at low intensity and high frequency, whereas increasing the K of the system is more convenient for AMFs that operate in highintensity and low frequency conditions [101]. Both strategies could be combined by creating magnetically coupled composites with two magnetic components [102]. The area of the high frequency hysteresis loops can be maximized by coupling a "soft" magnetic phase with the high MS and a "hard" magnetic phase with large coercivity through an exchange interaction [103]. Indeed, CoFe2O4@MnFe2O4 core-shell MNPs represent a paradigmatic example of this kind of system [104]. The synergetic effect of such magnetic coupling generates an increase in the SAR value of one order of magnitude with respect to similar MNPs with a single ferrite phase (3.03 kW/g). Even more impressive results were obtained with Zn0.4Fe2.6O4@CoFe2O4 core-shell MNPs prepared with a cubic shape in which the SAR was above 10 kW/g [105]. Although these are outstanding SAR values compared to other MNPs [106], the synthesis of homogeneous core-shell structures with a controlled shell thickness remains a challenge.

A more feasible approach to exchange coupling to develop advanced magnetic nanoheaters is the preparation of multicore MNPs, also known as nanoflowers. These structures are made of aggregates of magnetic nanocrystals with common epitaxial interfaces that couple the magnetic responses of the nanocrystals. Exchange coupling creates a cooperative magnetization process between the nanocrystals formed that increases the susceptibility of the aggregated nanostructures [107]. This increase in susceptibility implies a rise in the SAR values up to c.a. 2 kW/g at very high frequencies (700 kHz), even at moderate AMF intensities (25 kA/m) [108]. The amount of heat dissipated strongly depends on the primary crystal size, and on the composition and extent of the interfacial surface [109], creating an energetic balance of the dipolar interactions and exchange coupling between the primary nanocrystals that may favour or hamper heat dissipation [110].

Dipolar interactions are also a crucial parameter when using MNPs as nanoheating agents. When MNPs are magnetized they create their own dipolar field that affects the magnetic response of the surrounding MNPs. This effect is especially relevant for highly concentrated colloids and in aggregated systems in which the MNPs are in close proximity, such as endosomes. For small MNPs, the dipolar field may increase the anisotropy barriers and enhance their heating performance [111]. But in most cases the effect is the opposite, the dipolar field cancelling the effect of the applied field, reducing the susceptibility of the system and consequently diminishing their SAR [112,113]. This has been postulated as a possible cause for the discrepancies observed in the heating performance of MNPs between

ex vivo and in vivo studies [114]. Only in the case of elongated aggregates, such as chains, does the dipolar interaction result in cooperative magnetization that favours susceptibility and increases the coercivity of the system [115].

Although these parameters are interesting from the point of view of the magnetic properties, superparamagnetic iron oxide nanoparticles (SPIONs) are clearly the most advanced candidate in terms of commercial availability, regulation and clinical trials. In terms of magnetic properties, SPIONs present a relatively high susceptibility and low residual magnetization in the absence of an external magnetic field. Moreover, SPIONs are well-tolerated and they have low toxicity profiles, even in long-term studies [116]. In addition, they can be biotransformed from SPIONs to other iron compounds, facilitating their clearance [117–120]. SPIONs have been in clinical use for years, and several types have been approved for use in humans by the European Medicines Agency (EMA) and the Food and Drug Administration (FDA) in the USA, especially as anti-anaemic drugs and contrast agents for MRI [121].

#### *Preparation of Candidate MNPs for MHT*

In designing MNPs for HT, a balance must be reached between the size of the magnetic core to maximize the heat released (>10 nm) and the colloidal stability in biological media required for intravenous injection (<50 nm). Above these sizes, magnetic interactions between NPs are very strong and it is difficult to keep them apart despite their coating, such that they tend to aggregate and precipitate. One NP formulation already approved for use in cancer therapy is NanoTherm®, approved in 2010 by the EMA to treat recurrent GBM and in 2018 by the FDA for human prostate cancer (https://www.magforce.com/en/ home/about\_magforce/#highlights; accessed on 27 July 2021). NanoTherm® is a colloidal suspension of aminosilane coated 15 nm iron oxide NPs (with an iron concentration of 112 mg/mL) that can be delivered percutaneously into the tumour tissue. However, the challenge remains to develop NPs with enhanced specific loss of power and efficient delivery, within clinical AMF design constraints [122].

Iron oxide MNPs commercially available for HT are produced by precipitation of iron salts (Fe(II) and Fe(III)) in alkaline aqueous solutions. The size of the particles does not exceed 20 nm and it can be ensured by thermal treatment for long periods of time or by controlling the pressure in autoclaves. Thus, high pressure homogenisation processes allow the formation of individual crystals with mean diameters of 15–20 nm [123]. In a similar way, core–shell NPs and mixed ferrites are obtained by co-precipitation of stoichiometric mixtures of solutions containing divalent (Mn(II), Co(II), Zn(II)) and trivalent metals (Fe(III)) in alkaline medium [124]. Although core-shell MNPs can be finely tuned in organic media synthesis, the synthetic methods in aqueous media are preferred because they do not need additional coatings for water transference and are fully scalable to mass production, although controlling the size distribution and crystal order is limited due to the use of temperatures below 100 ◦C.

Controlling the shape of NPs and producing larger MNPs (>20 nm) can be achieved by thermal decomposition in organic media at temperatures as high as 300 ◦C, while cubes or rods are prepared using shape directing agents, either carboxylic acids or amines, respectively [96,97]. Autoclaves have recently been used to produce up to gram quantities of cubes, using iron pentacarbonyl as a precursor (Patent: WO2020222133A1). Other ferrites, such as manganese ferrite, zinc ferrite or a mixture of them in the form of core@shell or alloys prepared by this method, have showed excellent properties for HT [101].

Moreover, the assembly of magnetic cores into regular structures has been shown to significantly influence the HT behaviour of the particles, requiring the control of some key synthetic parameters that drive the self-assembly and growth process, such as surfactants and the viscosity of the medium. Thus, flower-like iron oxide assemblies between 25 and 250 nm can be obtained by heating a mixture of Fe(II) and Fe(III), or of an Fe(III) salt alone, to 200 ◦C in a heating mantel or in an autoclave [108,125]. In this sense, polyols are very interesting polar solvents that work as reducer, surfactant and high temperature synthesis

media, allowing the use of inorganic salts as precursors [126]. Moreover, it is possible to combine the polyol procedure with more efficient heating technologies, such as microwave heating, leading to higher production yields over shorter reaction times [127].

The assembly of magnetic cores can be exploited even in the absence of an interface between them. The chains of magnetic nanocubes naturally produced by magnetotactic bacteria have for long been the best candidates for MHT, with a SAR of 2.38 kW/g at 310 kHz and 30 kA/m [128]. For this reason, several mimetic systems have been produced using silica as a template or other anisometric nanomaterials [129]. Finally, further coating and functionalisation is possible for NPs obtained in either aqueous, organic and polyol media. Coating with aminosilanes and/or polysaccharides is mainly used for targeted HT applications [130]. These coatings provide an excellent first layer for the bioconjugation of biomolecules, such as antibodies or peptides, drugs or biomarkers.

#### **4. The Biological Effects of the Application of AFM to Cells Loaded with MNPs: Is It Always Hyperthermia?**

Having established the physical principles and the materials used in the development of MHT, we can focus on the biological effects that they produce. However, rather than adopting a general or macroscopic view of this issue, we shall focus on the cellular and subcellular effects of this treatment. There are many in vitro and in vivo studies that have correlated MNPs and AMF therapies with extended life expectancy or tumour regression [131], although the biochemical mechanisms responsible for such improvements remain unclear. Depending on the magnetic features of the MNPs, and on the AMF amplitude and frequency, the MNPs can transform the energy of the magnetic field into heat or mechanical effects (Figure 2c,d,e). The application of the AMF may cause the magnetic moments of the internalized MNPs to rotate in the direction of the field (Néel's relaxation) and the actual MNPs to physically rotate (Brown's relaxation). These two responses to the magnetic field are manifested to a greater or lesser extent depending on the intrinsic characteristics of the NMPs (size, shape, anisotropy, crystallinity) and the magnitude of the AMF. When the AMF is applied and it alternates at high frequency, the continuous re-orientation of the MNPs with the magnetic field alters the release of heat by the MNPs [132] or the physical movement of the MNPs, provoking mechanical damage.

Consequently, there is a controversy around if it could be termed hyperthermia for all effects provoked by MNPs + AMF treatment. So that, we propose to classify the biological effects into four groups, depending on the main cause of the biological effects observed. First, we talk about the most intuitive mechanism, the rise of temperature, that is, hyperthermia. Then, we mention several studies where the authors believe that the resulting effects are not related to temperature, but due to the physical and mechanical MNPs movements. Later, we introduce a section to propose an explanation for these apparently non-temperature-related effects, offering the possibility of a macroscopically undetectable hyperthermia phenomenon. Finally, we mention other indirect treatment processes that help us to tumour regression.

#### *4.1. Biological Effects of Heating*

It is well known that a rise in temperature triggers cell death, yet not all cell types are equally sensitive. Cancer cells are considered to be more susceptible to HT than healthy cells due to their higher metabolic rates [133], the hyperthermic inhibition of DNA repair [134] and the poorer heat dissipation through the blood flow [135]. More specifically, the biological effects of HT include (Figure 3): an increase in oxidative stress (Figure 3.<sup>1</sup> ) [136,137]; inactivation of membrane receptors and increase in ion permeability that affects cell transport (inhibition of amino acid transport and increased Na+, K+, and especially Ca2+: Figure 3.<sup>2</sup> ); a lack of stability and an increase in membrane fluidity (Figure 3.<sup>3</sup> ); changes in cytoskeletal organization, involving microtubule, microfilament and intermediate filament depolymerization (Figure 3.<sup>4</sup> ); increased protein denaturation and insoluble protein aggregation in the nucleus (Figure 3.<sup>5</sup> ), which promotes heat shock protein expression (Figure 3.<sup>6</sup> ) and centrosome damage, as well

as mitotic dysfunction (Figure 3.<sup>7</sup> ); and eventually, DNA damage or denaturation can occur (Figure 3.<sup>8</sup> ) [138].

**Figure 3.** Cellular and sub-cellular biological effects derived from hyperthermia. **Upper:** normal conditions (37 ◦C). **Bottom:** hyperthermia treatment (>42 ◦C).

#### *4.2. Biological Effects of Mechanical Rotation or Vibration*

It is interesting that there have recently been several reports of cell death after MHT without any perceptible rise in temperature [139–142]. The mechanisms responsible for these effects have not yet been elucidated, although there is data indicating that they are related to an increase in lysosomal permeability, which triggers an enhanced reactive oxygen species (ROS) production and enhanced activity of the lysosomal protease cathepsin D in the cytoplasm diminishing tumour cell viability [47,143,144]. This lysosomal permeabilization might be also caused by mechanical rotation or vibration of SPIONs altering lipid membrane stability (Figure 3.<sup>9</sup> ). Indeed, dynamic magnetic fields induce a slow rotation of lysosome targeted SPIONs, tearing the lysosomal membrane and activating apoptosis [145]. Moreover, according to theoretical simulations, the rotation of MNPs in a liquid media can be induced by either rotatory fields or AMFs. However, at moderate field intensities the torque induced by rotatory fields is 30 times higher than those created by

AMFs [61]. Evidence of the mechanical damage produced by MNPs under an AMF has come from reports of lysosomes rupture inside the cells [144].

#### *4.3. Biological Effects Derived from Non-Perceptible Heating: The "Hot Spot" Effect*

Another possible explanation for the biological effects of AMFs that are apparently unrelated to temperature changes might be very local intracellular heat release from SPIONs (not detected macroscopically), known as a "hot spot" effect. This local heatrelease enhances biological effects, such as the generation of ROS by the iron oxide surface of the NPs through the Fenton reaction (Figure 3.⑩), which is known to be accelerated directly by temperature [146]. Hence, we think that markers of sub-cellular temperature rises (e.g., Hsp70 [147] or thermal nanoprobes) should be implemented routinely in these studies to determine if tumour cell death can be always attributed to HT (even if these occur on a subcellular scale), or whether the biological effects that occur are unrelated to temperature.

#### *4.4. Biological Effects Derived from Other Indirect Process*

It seems clear that independently of whether they are due to MNP heating or their mechanical rotation and vibration, biological effects not directly or not exclusively related to temperature play a crucial role in tumour regression. ROS formation through Fenton reactions is probably the best studied of these, given that ROS can severely damage cell elements due to oxidation, such as DNA, proteins, lipids and enzyme cofactors, thereby inducing apoptosis [148]. ROS formation can occur by lysosome degradation or through the breakdown of other subcellular structures but also, by MNP interactions and Fenton reactions at the MNP's surface. Interestingly, and concomitant with ROS production, an increase in fluidity and a loss of cytoplasmic membrane integrity also activates cell death, either necrosis or apoptosis [5]. Other effects not directly related with the MNP-tumour cell interaction have an enormous importance in the fight against cancer. MHT has also been associated with activation of DCs and NK cells [149].

#### **5. MNP Behaviour in Response to AMF in the Biological Milieu**

In the previous section, we have analyzed the biological effects driven directly or indirectly by MNP exposure to AMF in MHT settings, that may or may not be trigger by MNPs heat release in response to AMF exposure. In recent years, many studies have suggested that the ability of MNPs to produce heat in response to AMF exposure when MNPs are in biological milieus is severely reduced, even sometimes undetectable, a circumstance that would not be desirable for MHT therapies. A possible explanation for this undesired behaviour could be that the magnetic response of MNPs to AMF was modified as a consequence of MNPs-cell interaction [150,151], being the causes of these changes in the magnetic properties a reduction in MNPs mobility, dipolar interactions, milieu viscosity, and MNPs clustering or aggregation [58,114,150–153]. This alteration of MNPs magnetic properties implies a dramatic reduction of SAR values, that could be observed when MNPs are aggregated by contact with cells, but also by contact with physiological milieus or viscous media emulating cellular environment, and depending on the intrinsic properties of the MNPs, these SAR decreasing values could be more than 60% [114,152], even in MNPs that, after being tested in aqueous medium, showed a promising heating capacity [153]. Most of the observed alterations in the MNPs magnetic properties could be explained by the restriction of the Brownian relaxation, since MNPs cannot respond to AMF with rotation because MNPs are physically immobilized and blocked. Therefore, when Brownian relaxation component is suppressed in biological conditions, and it was demonstrated that Néel relaxation was unaffected by changes to their biological microenvironment, emphasizing the importance of MNP intrinsic magnetic properties for MHT when particle mobility cannot be kept. So that, Néel relaxation component becomes the only possible heat induction mechanism [58]. Likewise, doping MNPs with Zn allows a

strong Néel relaxation that was preserved after MNPs-cell interaction, which is suitable for heat releasing in MHT [151].

Although Néel relaxation contributes for heating generation during MHT the most important component for heating during an ideal MHT is the Brownian relaxation. Therefore, it is a key point to understand why the MNPs are immobilized in and if it is a reversible process. It has been shown that MNPs that were blocked as a consequence of cell internalization, can recover their original magnetic properties, including the Brownian relaxation, when the cells that contained them were lysed, due to the integrity of the magnetic core is preserved during this process [57] suggesting that MNPs immobilization or aggregation is the final cause of loss of Brownian relaxation. In another study, three systems with MNPs different spatial distributions and grades of aggregation were analyzed in order to compare their magnetic properties: isolated MNPs, MNPs-liposomes system, and MNPs-cell interaction (using, in turn, Jurkat cell line that attached MNPs to the outer membrane and Pan02 cell line that internalized the MNPs). Results showed that the biological environment played a crucial role in the dynamic magnetic response of the MNPs, being more altered for MNPs-cell system, and concluding that the simple fact of being in contact with the cells triggers MNPs aggregation [154].

Other studies tried to determine if this aggregation derived from MNPs-cell contact was a process dependent on the intrinsic properties of the MNPs or the host cell line. For that, co-precipitated maghemite nanoparticles, assembly of the same maghemite nanoparticles in liposomes, cobalt ferrite nanoparticles, iron oxide/gold dimers, iron oxide nanocubes and iron oxide nanoflowers were tested in three different biological environments: MNPs in water, MNPs attached to adenocarcinoma SKOV-3 cells membranes, or MNPs internalized in SKOV-3 cells. As result, a rapid fall in the heating capacity of all the nanomaterials tested (regardless of its different composition, shape or size) has been observed when MNPs were associated with the cell membrane or were internalized [153]. Likewise, different core size MNPs (6, 8 and 14 nm), coating (APS: 3-aminopropyltriethoxysilane, and DMSA: dimercaptosuccinic acid), cell line (Jurkat and Pan02) and subcellular localization (membrane or internalized in endosomes/lysosomes) were tested in biological milieus being demonstrated that the aggregation process was independent of MNPs core size, coating, cellular environment, host cell line and MNPs subcellular localization [150].

Currently, these problems of aggregation or blockade of MNPs are the main bottlenecks in MHT therapies, so that more efforts will be required to develop strategies directed to avoid them to achieve a satisfactory MHT.

#### **6. Rational Design of Strategies Based on MNPs for MHT and Their Applications to Tumours** *In Vitro* **and** *In Vivo*

Having described the physical and biological behaviour of MNPs in relation to MHT, we shall assess how this knowledge has been directed towards extracting the full potential of this therapy through the rational design of MHT strategies based on MNPs. Hence, it is important to consider the different intrinsic (size [155], shape [156], doping, etc ... ) and extrinsic (magnetic field intensity and frequency, subcellular location, intracellular aggregation, etc ... ) parameters that govern the success of such treatments (Scheme 1). As such, the different studies where MHT has been successfully achieved based on the rational design of *in vitro* and *in vivo* experiments will be considered.

**Scheme 1.** Main parameters to be considered to achieve a successfully magnetic hyperthermia treatment.

#### *6.1. Fine-Tuning the Intrinsic and Extrinsic MNP Properties for In Vitro Magnetic Hyperthermia*

We can find many examples of rationalized MNP designs that have been considered for in vitro MHT. For instance, size-optimized MNPs enhance cell death in Jurkat cells after MHT [157]. Likewise, phospholipid-PEG coating has been used to concurrently deliver Doxorubicin and to generate heat for an enhanced multimodal cancer treatment in HeLa cells [158]. Likewise, MNP functionalization with the folate-receptor (a tumour marker) has been employed for smart delivery to the MCF7 and G1 cell lines, with no uptake by a control L929 cell line [159]. Furthermore, it has been described that MNPs pegylation coating is capable of counteracting the interactions between dipolar particles while maintaining a low level of nanoparticle aggregation in environments of different ionic strengths and viscosities [160]. Each of these serves as a good example of the rationalization of MNP size, coating and targeting for in vitro MHT.

We carried out studies that combined different strategies for the rational design of MNPs [161], synthesizing them by thermal decomposition to obtain 18 nm flowerlike Mn-doped SPIONs covered with DMSA and functionalized with cRGD (an αvβ3- Integrin-Ligand) peptide (from now on named NF-DMSA-PEP) that targets the U87MG glioblastoma cell line. These NF-DMSA-PEP had higher SAR values than 12 nm spherical MNPs covered with DMSA (NP-REF) and 20 nm flower-like MNPs covered with citric acid (NF-REF), demonstrating the notable role of rationalized intrinsic MNPs properties (size, shape, doping and coating MNPs). Furthermore, extrinsic properties related to the biological features of the target cells must also be considered. MNPs uptake was tested by comparing our NF-DMSA-PEP system with the same MNPs without the cRGD peptide (NF-DMSA). NF-DMSA-PEP uptake by U87MG cells was enhanced 5–6 fold, while endocytosisexocytosis cycles avoided compact aggregation inside lysosomes and the resulting decrease in NP-induced HT. Consequently, peptide effectiveness was demonstrated, establishing 2 h as the optimal time to then apply the magnetic field. Finally, a 2 h NF-DMSA-PEP incubation followed by a 1 h AMF application (25 kA/m, 250 kHz) efficiently induced intracellular cell heating (Hsp70 over-expression), ROS production and cell death (but

without inducing apoptosis). The biological effects observed were always stronger with NF-DMSA-PEP than with control MNPs.

#### *6.2. Tuning Intrinsic and Extrinsic MNP Properties for In Vivo Magnetic Hyperthermia*

One further step in the study of these approaches is the application of rational design to treat tumours in animal models, mainly glioblastoma, pancreas, breast and prostate. The most common way of administering nanoparticles for antitumor hyperthermia treatments is by intratumoral injection of MNPs, with or without the aid of the use of advanced imaging techniques to deliver MNPs into the tumour [162,163]. Systemic administration of MNP through intravenous injection followed by the biological or physical targeting of those MNPs to the tumour is another possibility, especially in the case of hard-to-reach tumors. However, MNPs doses needed is greater than for intratumoral administration, because the amount of MNPs that reach the tumour depend on several factors such as biodistribution [51], EPR effect [31], active targeting [33,34] and renal clearance [31]. These factors must to be taken account during the rational design process of the MNPs for MHT. For instance, the pegylation of MNPs allows long blood circulation times avoiding the rapid uptake by mononuclear phagocytic system and renal clearance [164]. Moreover, thinking in translational therapy, several studies conclude that EPR effect work properly in rodents but not in humans [32] making essential the use of active targeting strategies that could make the process more complex. Nevertheless, satisfactory MHT through rational designed MNPs intravenously administered has been done, as exemplified a study where a rational design based on pegylation of MNPs and functionalization with c(RGDyK) peptide solved the renal clearance and active targeting issues respectively, and finally glioblastoma regression in mice was achieved [165].

For intratumoral administration rational MNPs design is also needed to improve the regression of tumours. For instance, MNPs covered with DMSA and functionalized with doxorubicin as chemotherapeutic agents were administered intratumorally achieving breast cancer xenograft regression through a synergic effect [166]. Likewise, the development of a biocompatible magnetic lipid nanocomposite vehicle for encapsulate MNPs and doxorubicin was demonstrated to provoke a synergic effect sensitizing the tumour cells to cancer chemotherapy in a subcutaneous melanoma mice model [167]. Another example is the design of Janus MNPs charged with doxorubicin achieving the decrease of tumour weight in subcutaneous breast solid tumour models [168]. Another example, cubic-shaped MNPs, since cubic-shaped MNPs are better heaters than spherical MNPs [97], coated individually with a polymer shell to avoid MNPs aggregation, have been shown to be effective heat mediators for MH and heat-mediated chemotherapy on an in vivo xenograft tumour model using A431 epidermoid carcinoma cells [169].

In addition to the intrinsic properties of MNPs, the extrinsic properties must be considered. It was shown above that controlling and optimizing the biological parameters related to the MNP-cell interactions is important for satisfactory MHT. However, since intra-tumour injection is normally chosen in vivo, it is more interesting to consider other parameters like the animal model or the type, size and location of the tumour. In addition, optimizing the HT equipment conditions will play an even more crucial role. Modifying the magnetic field frequency and intensity will allow the MNP induced heating capacity be finetuned, with a higher frequency and a stronger magnetic field intensity translated into more heating, which is crucial for certain nanomaterials. Normally, MNPs require a threshold field to open their hysteresis loop and then achieve heat release. Nevertheless, higher magnetic field frequencies and intensities are not permitted for in vivo or translational therapies and thus, a compromise between these parameters must be found in order to remain within the safety limits [79–81]. The duration and the repetition of applications are other parameters to be considered. A full comparison of the in vitro and in vivo conditions for MHT has recently been prepared by Vilas-Boas et al. [170].

#### **7. From the Laboratory to the Clinic**

Despite the improvements implemented and the exponential growth of studies into MHT to treat cancer, translational investigation in this area has not progressed as desired and its clinical implementation has not yet occurred [79]. Various factors have influenced this delay, technological challenges being the most important. As mentioned previously, only one NP formulation has currently been approved for HT: NanoTherm® (MagForce AG, Berlin, Germany). This sole alternative rules out the possibility employing a rational design of NPs with better physical characteristics and a potentially improved heating capacity at the same concentrations or doses [171]. In addition, NanoTherm® is a ferrofluid that agglomerates in the tissue, and entrapment by macrophages as opposed to glioblastoma cells might result in cancer cells receiving insufficient doses [172]. Moreover, intratumour NPs injection must be employed to avoid aggregation, restricting their use to solid and accessible tumours like glioblastoma or prostate cancer. Consequently, a homogeneous NP distribution and therefore, constant heat distribution across the tumour is difficult to achieve [171]. Moreover, the inability to use active rational design also translates into a loss of effectiveness. On the other hand, the AMF applicator MFH 300F® (later implemented as Nanoactivator® F100: MagForce s AG, Berlin, Germany) is the only apparatus used in the clinical trials carried out to date, and always operating at a fixed frequency of 100 kHz and with a field strength of 0–18 kA/s [79]. However, changes to these parameters might produce better results while still respecting the safety limits.

Several clinical trials have been perfromed by Charité—Universitätsmedizin Berlin (Germany) and the spin-off company MagForce AG, Berlin (Germany). The pipeline that has driven the initial idea towards the realization of clinical trials can been easily traced (Figure 4). In 1993, the potential use of SPIONs for HT therapy was first noted and how NP application localised to a tumour might be less invasive than other techniques. With clinical implementation in mind, a moderate concentration of ferrite 5 mg/g tumour was considered, coupled to clinically acceptable magnetic fields that were comparable to radiofrequency heating by local application and superior to regional RF heating [173]. Nearly a decade later, in 2001 this idea was developed further and a new magnetic field therapy system was introduced for the treatment of human solid tumours with magnetic fluid HT. In this study, two of the three pillars of HT therapy used in the Charité-MagForce clinical trials were well defined. Firstly, the aminosilane magnetite NPs used were subsequently manufactured by MagForce AG and called NanoTherm® and they were seen to be more significantly taken-up by malignant cells than normal cells. Secondly, the first prototype of a magnetic fluid hyperthermia (MFH) therapy system was designed (Applicator MFH 300F) and which was later developed as Nanoactivator® F100 by MagForce AG [174,175]. Finally, and just one year later the third pillar appeared, the software initially called HyperPlan and now NanoPlan® (MagForce AG) that enables treatments to be planned through a thin-sliced CT or MRI scan. The software developed, in combination with the AMIRA® visualization package (Mercury Computer Systems, Berlin, Germany), allows us to obtain 3D reconstructions of the NP distribution in the tumour and the localization of the thermometry catheter. Moreover, the physician can modify the parameters to simulate different scenarios and to determine the optimal magnetic field strength for the treatment, estimating the possible temperature distribution during the treatment [176].

Once the basic concepts have been fixed and a well-defined route obtained, a pilot clinical trial was carried out. The aim of this pilot study was to evaluate whether the MFH technique can be used for minimally invasive treatment of prostate cancer. The results indicated that HT using MNPs injected transperineally into the prostate was feasible and well-tolerated. Moreover, NPs were retained for at least 6 weeks in the prostate, making sequential HT treatment possible without the need for new NP application. This study formed the basis on which future clinical trials could be designed [177].

**Figure 4.** Timeline of the major milestones and clinical trials carried out on humans where Magnetic Hyperthermia Treatment was employed.

In the following years different clinical trials were performed, the details of which are shown in Table 2. These clinical trials focused on understanding, optimizing and improving particular aspects of the technique to enhance the results. First, in 2006 a phase I trial was performed to evaluate the feasibility and tolerability of thermotherapy using MNPs in different pre-treated tumours, as well as testing three different NP injection methods. The results showed that magnetic fluid and thermotherapy treatment was well-tolerated, with no or only moderate side-effects, respectively. Moreover, there was a clear need to further improve the temperature distribution by refining the implantation techniques, or simply by increasing the amount of NPs or the magnetic field strength [178].

A year later, a phase I trial was carried out in 2007 to investigate the feasibility of using thermotherapy with biocompatible SPIONs in patients with locally recurrent prostate cancer, evaluating an image-based approach for the non-invasive calculation of the 3D temperature distribution. It was concluded that heating using MNPs was feasible. Hyperthermic to thermoablative temperatures were achieved in the prostate at 25% of the available magnetic field strength, indicating the potential to reach higher temperatures and that a specific non-invasive thermometry method could be developed that may be used for thermal dosimetry [179]. In the same clinical trial, the treatment-related morbidity and quality of life (QoL) during thermotherapy was studied, and it appeared that interstitial heating using MNPs was feasible and well-tolerated by patients with locally recurrent prostate cancer. Furthermore, deposition of NPs in the prostate was evident 1 year later, even though a homogeneous distribution was not achieved. Finally, a refinement of the technique was needed to enable higher magnetic field strengths to be applied [180].

In parallel, a phase I trial was presented in 2007 to evaluate the feasibility and tolerability of the newly developed thermotherapy, using MNPs to treat recurrent GBM and guiding the intratumour NP injection by 3D imaging. The study demonstrated that thermotherapy using MNPs is safe to use in the treatment of brain tumours and that therapeutic temperatures ranging from HT to thermoablation can be achieved. These promising results opened the door to further studies [181] and consequently, a few years later a phase II clinical trial was carried out on 59 GBM patients. The objective of the study was to determine the efficacy of intratumour thermotherapy in conjunction with fractionated stereotactic radiotherapy for glioblastoma. This clinical trial confirmed the aforementioned advantages, demonstrating MHT to be a safe and well-tolerated cancer therapy. Moreover, the clinical trial concluded that the combination of HT and radiotherapy was clinically effective, augmenting the overall survival of patients. Finally, it was proposed that the combination of HT and chemotherapy (particularly temozolomide), in conjunction with intratumour NP targeting using convection-enhanced delivery (CED) is a promising approach for the treatment of other solid tumours that should be evaluated in future clinical trials [59].


**Table 2.** Summary of the different MHT clinical trials carried out by Charité –MagForce.

Finally, the "Magnetic Nanoparticle Thermoablation-Retention and Maintenance in the Prostate: A Phase 0 Study in Men (MAGNABLATE I)" was the first clinical trial performed outside of the Charité –MagForce collaboration, promoted by University College London Hospitals (NCT02033447 [182].

#### **8. Conclusions**

From this review we can draw some important conclusions regarding the clinical implementation of MHT. In the case of MNPs, there are many studies about how their intrinsic properties (size, shape, composition, coating, etc) can affect their ability to generate heat in response to AMFs in non-biological milieus. This knowledge can be used to improve their rational design for MHT therapies to obtain the better heat release. However, the behaviour of MNPs in response to AMF in biological media, and the underlying cellular mechanisms that are triggered, are not yet fully understood, representing a major bottleneck in the application of MHT under physiological conditions. Therefore, it is important that more studies focus on the behaviour of MNPs inside cells. For example, it is important to study MNP biodegradation in different cell types and tissues, since MNP degradation in cells could affect their magnetic properties during treatment, and issue that could be critical in therapies that require the repeated application of the magnetic pulses long after the administration of NPs. In terms of the development of HT equipment, it is important to emphasize the importance of selecting the appropriate field intensity and frequency for each case, so that treatments produce specific effects. For clinical applications, it should be remembered that safety limits exist above which we cannot operate.

In general terms, the clinical implementation of MHT has not progressed as might have been anticipated, despite the success of some clinical trials. The global conclusions that we can extract from the clinical trials carried out to date is that the first steps taken have yielded promising results. However, in recent years no more clinical trials have been carried out. Fortunately, this tendency seems to be changing and a few months ago, a new and very ambitious pivotal single-arm clinical study for the focal ablation of intermediaterisk prostate cancer using NanoTherm® was presented by MagForce USA, Inc. following its FDA approval. Along similar lines, in Europe both the Vall d'Hebron University Hospital and the Fuenlabrada University Hospital are involved in a new feasibility study on treating locally advanced pancreatic ductal adenocarcinoma (PDAC) as part of the NoCanTher project [183].

In summary, we strongly believe that increasing knowledge in the key biological aspects above-mentioned is highly necessary to achieve fine control of the process that could trigger the desired clinical implementation of MNPs-based magnetic hyperthermia.

**Author Contributions:** D.E.-B., J.G.O., M.d.P.M. and D.F.B. conceived and designed the review. D.E.-B. and J.G.O. wrote sections of the manuscript, and prepared the figures and tables. M.d.P.M. and D.F.B. coordinated, critically revised, modified and completed the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** DEB is a pre-doctoral scholar working under a FPI Contract (PRE2018-084189) from the Spanish Ministry of Science and Innovation. This work was supported in part by grants from the Spanish Ministry of Science and Innovation (SAF-2017-82223-R and PID-2020-112685RB-100 to DFB, and MAT2017-88148-R, to MPM) and by the European Commission through the HOTZYMES project (H2020-FETOPEN-RIA 829162 to MPM). The groups of DFB and MPM are part of the Network "Nanotechnology in Translational Hyperthermia" (HIPERNANO, RED2018-102626-T) supported by the Spanish Ministry of Science and Innovation.

**Acknowledgments:** We thank members of the laboratories led by Domingo F Barber (CNB-CSIC, Madrid) and María del Puerto Morales (ICMM-CSIC, Madrid) for their helpful comments and discussions. The authors are also grateful to Mark Sefton (BiomedRed SL) for English language editing of the manuscript.

**Conflicts of Interest:** The authors have no conflicts of interest to declare.

#### **References**


## *Review* **Magnetic Particle Imaging: An Emerging Modality with Prospects in Diagnosis, Targeting and Therapy of Cancer**

**Zhi Wei Tay 1,\*, Prashant Chandrasekharan 2, Benjamin D. Fellows 2, Irati Rodrigo Arrizabalaga 2, Elaine Yu 2, Malini Olivo <sup>1</sup> and Steven M. Conolly <sup>2</sup>**


**Simple Summary:** Magnetic Particle Imaging (MPI) is an emerging imaging technique that provides quantitative direct imaging of superparamagnetic iron oxide nanoparticles. In the last decade, MPI has shown great prospects as one of the magnetic methods other than Magnetic Resonance Imaging with applications covering cancer diagnosis, targeting enhancement, actuating cancer therapy, and post-therapy monitoring. Working on different physical principles from Magnetic Resonance Imaging, MPI benefits from ideal image contrast with zero background tissue signal, enabling hotspot-type images similar to Nuclear Medicine scans but using magnetic agents rather than radiotracers. In this review, we discussed the relevance of MPI to cancer diagnostics and image-guided therapy as well as recent progress to clinical translation.

**Abstract:** Background: Magnetic Particle Imaging (MPI) is an emerging imaging modality for quantitative direct imaging of superparamagnetic iron oxide nanoparticles (SPION or SPIO). With different physics from MRI, MPI benefits from ideal image contrast with zero background tissue signal. This enables clear visualization of cancer with image characteristics similar to PET or SPECT, but using radiation-free magnetic nanoparticles instead, with infinite-duration reporter persistence in vivo. MPI for cancer imaging: demonstrated months of quantitative imaging of the cancer-related immune response with in situ SPION-labelling of immune cells (e.g., neutrophils, CAR T-cells). Because MPI suffers absolutely no susceptibility artifacts in the lung, immuno-MPI could soon provide completely noninvasive early-stage diagnosis and treatment monitoring of lung cancers. MPI for magnetic steering: MPI gradients are ~150 × stronger than MRI, enabling remote magnetic steering of magneto-aerosol, nanoparticles, and catheter tips, enhancing therapeutic delivery by magnetic means. MPI for precision therapy: gradients enable focusing of magnetic hyperthermia and magnetic-actuated drug release with up to 2 mm precision. The extent of drug release from the magnetic nanocarrier can be quantitatively monitored by MPI of SPION's MPS spectral changes within the nanocarrier. Conclusion: MPI is a promising new magnetic modality spanning cancer imaging to guided-therapy.

**Keywords:** magnetic particle imaging; magnetic nanoparticles; magnetic hyperthermia; magnetic drug delivery

#### **1. Introduction**

Magnetic Particle Imaging (MPI) is an emerging magnetics-based imaging technique first introduced by Philips, Hamburg in 2005 [1]. While the name is very similar to Magnetic Resonance Imaging (MRI), it operates on very different physical principles. Unlike MRI, where the signal comes from the precession of nuclear spin magnetic moments of the target nuclei (e.g., 1H, 2H, 13C, 17O, 19F, 23Na, 31P), the MPI signal is obtained from

**Citation:** Tay, Z.W.; Chandrasekharan, P.; Fellows, B.D.; Arrizabalaga, I.R.; Yu, E.; Olivo, M.; Conolly, S.M. Magnetic Particle Imaging: An Emerging Modality with Prospects in Diagnosis, Targeting and Therapy of Cancer. *Cancers* **2021**, *13*, 5285. https://doi.org/10.3390/ cancers13215285

Academic Editors: Moriaki Kusakabe and Akihiro Kuwahata

Received: 31 July 2021 Accepted: 19 October 2021 Published: 21 October 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

the ensemble magnetization of superparamagnetic iron oxide nanoparticles (SPION) as described by the Langevin model [2]. Because there are no SPIONs found in native biological tissue unlike the 1H in water and biological tissue sensed by MRI, MPI benefits from zero tissue background signal and achieves excellent image contrast comparable to tracer images typical of nuclear medicine scans such as positron emission tomography (PET) or single-photon emission computerized tomography (SPECT), which are the gold standard for diagnostic cancer imaging [3–5]. Since only SPIONs produce signal in an MPI scan, the MPI images obtained are fully quantitative in a linear fashion and are robust to minute changes in susceptibility. In comparison, the same SPIONs in an MRI scan are typically semi-quantitative as they produce contrast changes via susceptibility differences (Figure 1a), yielding a non-linear indirect effect on the 1H signal [2]. MPI operates in the kilohertz frequency range where magnetic fields fully penetrate tissue, bone, and air with negligible attenuation and reflection differences. Thus, MPI does not have any view limitations and works robustly even in lungs [6–9] and bones, which are challenging for MRI [2,10] and ultrasound.

**Figure 1.** Overview figure for Magnetic Particle Imaging's prospects in diagnosis, targeting, and therapy of cancer. Magnetic Particle Imaging (MPI) is an emerging magnetic imaging technology that works completely differently from MRI, providing radiation-free tracer-like contrast and linear quantitation with nanogram sensitivity to superparamagnetic iron-oxide nanoparticles (SPIONs). While most MPI research is still preclinical, MPI hardware has recently reached clinical scale scanners and is en route to clinical translation. (**a**) Zheng et al. 2016 [8]—Comparison of MRI SPION contrast to MPI contrast of the same SPIONs. (**b**) Arami et al. 2017 [11] — schematic of cancer-targeted magnetic nanoparticles. (**c**) Yu et al. 2017 [12]—MPI image of SPIONs accumulated within a tumor. (**d**) Banura et al. 2017 [13]—magnetic steering of magneto-aerosol with MPI as post-event verification. (**e**) Dames et al. 2007 [14]—magnetic tip for targeted delivery of magnetic

aerosol to lung. (**f**) Tay et al. 2018 [15]—MPI scanner's gradients enable pinpoint heating at user-selected locations, heating SPIONs at the tumor without heating off-target SPIONs. (**g**) Liu et al. 2018 [16] — magnetic gradients enable pinpoint drug release at target location without triggering release from adjacent nanocarriers 2 mm away. (**h**) Zhu et al. 2019 [17] — MPI in vivo non-invasive quantification of the percentage of release of drug from nanocarriers in mice tumors, enabling real-time assessment of the success of drug delivery for cancer. Figures within insets reproduced with permission from respective authors cited in reference numbers and publishers.

> Besides the excellent image contrast, one of the other key benefits of Magnetic Particle Imaging for cancer imaging is the relatively high sensitivity of the modality. The electronic magnetization of SPIONs sensed by MPI is 22 million times stronger than that of the nuclear magnetization of water (1H) at 7 Tesla [2]. Furthermore, the dose limit of iron oxide is 510 mg according to Lu et al. 2010, which is 25 million times higher than the 2 ng dose limit of PET [18,19]. The 2 ng value was calculated from the 370 MBq FDA-mandated dose limit for 18-FDG divided by the specific (radio)activity of 1ng of 18-fluorine (in MBq, averaged specific activity value). This implies that MPI can increase the systemic administered dosage to compensate and ensure tumor detection at 260 nM Fe sensitivity, even though this is poorer sensitivity than the 2 pM of PET [20]. Therefore, if SPIONs can achieve similar targeting efficiencies to tumors as 18-FDG, MPI can be expected to be competitive with PET on a dose-limited comparison, and thus help avoid radiation dose (especially important for pediatrics). Other advantages include the near-infinite signal half-life of SPIONs enabling longer time for circulation and binding to tumors while the 110 min half-life of 18-FDG necessitates a PET scan merely 40 min after injection [19]. Convenience is also improved as SPIONs can be used off-the-shelf, thus avoiding cyclotron facility overheads and the radiation safety measures for hot chemistry preparations.

> Regarding the imaging agent, the SPION magnetic "tracers" used in MPI will differ depending on the application. For stem cell or immune cell labeling, carboxydextran-based SPIONs have shown good labeling efficiency, likely due to the affinity of the dextran coating to cell uptake and internalization. For vascular imaging or tumor targeting, longcirculating stealth SPIONs with PEG-based coatings are ideal due to the enhancement of blood circulation half-life, allowing more time for the tracer to remain in circulation for vascular imaging or for the tracer to aggregate into the tumors. For magnetic hyperthermia applications where the magnetic nanoparticles are heated via external alternating magnetic fields (AMF), the magnetic core of the SPIONs must demonstrate high heating performance, i.e., good specific absorption rate (SAR) values at typical hyperthermia frequencies (100–1000 kHz) [21]. It is important to note that these qualities are not mutually exclusive and an SPION can be designed with multiple of these qualities such as a high heating performance magnetic core with stealth PEG-coating.

#### **2. Physical Mechanisms Underlying Magnetic Particle Imaging**

In brief, MPI performs spatial encoding, signal detection, and image reconstruction based on very different magnetic principles from MRI. From the high-contrast and spatial resolution characteristics, MPI is more similar to PET and SPECT, although it uses nonradioactive SPIONs at 20–100 nm sizes rather than small-molecule radiotracers. This section will explain the mechanism of MPI and its spatial and temporal resolution.

#### *2.1. Localization and Collection of Signal from a Specific Slice or Volume*

Magnetic Particle Imaging has two methods of localization of signal and thereby achieves spatial encoding in order to reconstruct an image. For the system matrix method, a point source of SPIONs is physically moved to every voxel in the field-of-view (FOV) and the MPI harmonic signal recorded as a calibration to determine the system matrix transfer function. In order to encode a different MPI harmonic signal at every voxel, a static background selection field is applied. The selection field is defined as a gradient field (magnetic field strength varies spatially) with a zero-field region at a central point (defined as field-free-point system—FFP) or a zero-field in a line geometry (defined as field-freeline system—FFL) as shown in Figure 2a,b. The differing background field strength as a function of position changes the SPION magnetization and results in a different MPI harmonic signal depending on position in the FOV. The static selection field alone cannot excite an MPI signal, and thus a time-varying drive field of 25 kHz and 16 mT is applied. The definition of a drive field is a monotonal excitation magnetic field operating in the kilohertz frequency range that aims to generate rapid magnetization changes in SPIONs as the MPI signal. When superimposed on top of the selection field, the result is the motion of the FFP in a Lissajous trajectory so as to pass near every voxel in the FOV at least once during the scan. In conclusion, spatial encoding for the system matrix meth-od uses the combination of the selection field gradient and a Lissajous (rather than raster) trajectory to determine a unique MPI harmonic signature for each and every voxel in the FOV [5].

**Figure 2.** Physical mechanisms underlying how Magnetic Particle Imaging (MPI) scans and produces an image. (**a**) Scanner configuration for a field-free-line (FFL) MPI scanner. (**b**) Scanner configuration for a field-free-point (FFP) scanner. (**c**) Illustrative workflow diagram on how the previously defined MPI-related magnetic fields can be applied to SPIONs that have accumulated within a tumor in terms of how to spatially encode, signal detect, and image reconstruct the MPI image. The background tumor image is adapted from Jhaveri AM, Torchilin VP. Multifunctional polymeric micelles for delivery of drugs and siRNA. Front Pharmacol 2014; 5:77 under a CC by 3.0 creative commons license [22].

For the x-space method, there is no pre-calibration step. Spatial encoding relies on the fact that SPIONs physically at the location of the FFP or FFL give the largest amount of signal and SPIONs away from the FFP give less and less signal in accordance to the point-spread-function (PSF) of that specific MPI scanner and SPION combination. This is basically predicated on the PSF as imaging spot size, with a smaller spot size enabling greater precision in discerning at-FFP signal and suppressing off-FFP signal. To acquire the entire FOV, the FFP or FFL is rastered across the FOV, usually in a cartesian trajectory, although the Lissajous trajectory used in the system matrix method could work too. The x-space drive field is usually in the same-axis as the detector coil (single-axis) and uses typical values of 20 kHz and 20 mT [3].

#### *2.2. Signal Detection and Image Reconstruction Approach for MPI*

MPI uses receive coils, defined as inductive solenoid or saddle-shaped wire coil sensors, which detect the SPION signal based off the time-varying magnetization changes of the SPION in response to the drive field. As a result, the signal strength is proportional to dM/dt and frequencies around 20 kHz are preferred as a trade-off between high signal strength from dM/dt and minimizing relaxation-induced blurring when the SPION cannot keep up with the drive field switching [3]. Unlike MRI where the readout timing is usually delayed after the excitation, the MPI signal is read out at the same time as the drive field (excitation) application. There is thus a large amount of direct feedthrough of the drive field into the inductive received signal. This is mitigated by high-pass or band-pass filters as well as gradiometric sensing coil design.

Image reconstruction for the system matrix method solves an inverse problem using the calibrated system function (3D matrix) achieved by recording the MPI harmonics at each and every voxel in the FOV. Image reconstruction for the x-space method relies on knowledge of the instantaneous position of the FFP and FFL in 3D space. The current instantaneous MPI signal is directly gridded to the known FFP/FFL location [3].

#### *2.3. Spatial Resolution and Time Requirements for MPI*

The current spatial resolution for MPI is around 0.5–2 mm depending on the magnetization characteristics of the SPION used as well as the gradient strength of the scanner [2]. Figure 2 compares the resolution and sensitivity of MPI to other imaging modalities. The temporal resolution of MPI can be relatively good at 45 frames per second as achieved by system matrix MPI due to the speed of the Lissajous trajectory [5].

#### **3. Imaging Cancer Using Magnetic Particle Imaging**

In brief, MPI is similar in image-quality to PET because of its zero-background, highcontrast, and ~1 mm spatial resolution. However, MPI's imaging agent of SPIONs cannot utilize the Warburg effect to target tumors and must rely on other mechanisms such as targeting of cancer cell receptors or cancer-specific proteases. This section discusses the various imaging studies on cancer that have been performed with MPI.

The gold standard for clinical cancer imaging is Positron Emission Tomography (PET). The ability of 18-FDG to selectively accumulate in even small metastatic tumors due to the Warburg effect and the tracer-nature and positive contrast of PET scans allow for unambiguous diagnosis of tumor presence and location [20]. Coverage of the whole-body is possible except for the brain or bladder due to the low contrast caused by high background 18-FDG uptake by healthy tissue. However, PET scans still have a non-negligible radiation dose and are not recommended for pediatric imaging. MRI and CT are also widely used for cancer imaging but often require tumors to be relatively large for reliable detection on scans. In this context, Magnetic Particle Imaging is promising as it provides tracer-like contrast (see Figure 1a–c) without any radiation dose due to the use of magnetic "tracers" as opposed to radiotracers. Figure 3 summarizes the differences of MPI from other imaging modalities for cancer imaging. In practice however, the SPIONs used in MPI still need to

make progress towards matching 18-FDG's high affinity to cancerous tissue in order to be competitive with PET.

**Figure 3.** Comparison of MPI to other molecular imaging modalities. Figure on right panel adapted with permission from Saritas et al. J. Magn. Reson. 229 [2]. Copyright 2013 Elsevier.

The earliest demonstrated application of Magnetic Particle Imaging towards cancer was a 2014 in vitro study by the University Hospital of Schleswig Holstein and Institute of Medical Engineering at the University of Luebeck, Germany [23]. Custom dextran-coated SPIONs (UL-D) were synthesized and demonstrated good internalization by head and neck squamous cancer cells as well as significant MPI signal via in vitro measurements using a Magnetic Particle Spectrometer. Although in vivo images were not demonstrated, the authors comprehensively characterized the labeled cells showing that their MPI-suitable SPIONs did not impact cell mitochondrial activity (MTT assay), cell viability (annexin V-APC-Propidium Iodide flow cytometry), cell proliferation (xCELLigence DP), cytokine secretion (Bead-based immunoassays for IL-6, IL-8, IL-1β and TNF-α), and reactive oxygen species generation (ROS assay by Dichlorofluorescein diacetate). These assays suggest that labeling of the cancer cells should not negatively impact tumor behavior such as increased tumor invasion or metastases.

In 2016, another in vitro study demonstrated the detection of cancer-specific proteases using changes in Magnetic Particle Spectrum (MPS) of MPI-compatible monodisperse iron oxide nanoparticles [24]. The linker-peptide-aggregated nanoparticles demonstrated a significant change in their spectrum when exposed to cancer-specific proteases. Although this assay was not verified for in vivo MPI, since MPI can be calibrated to tune specifically to a designated MPS (color MPI), this strategy could be promising to increase MPI imagespecificity to cancer cells. Sensitivity can be improved by optimization of the magnetic core size [25] as well as designing contrast-enhancing MPI pulsed excitation rather than continuous-wave excitation [26].

The earliest full study of in vivo Magnetic Particle Imaging of cancer (Figure 4a) was demonstrated in 2017 by the University of California Berkeley on their academic MPI scanner using long-circulating SPIONs (LS-008) from Lodespin Laboratories [12].

**Figure 4.** Imaging Cancer using Magnetic Particle Imaging. Figures adapted with permission from Yu et al. Nano Lett. 17. Copyright 2017 American Chemical Society. (**a**) Yu et al. 2017 [12] used LS-008 SPIONs injected i.v. at 15 mg/kg into an MDA-MD-231-luc flank xenograft in rat. The long-circulating SPIONs were non-targeted and after a few hours of systemic circulation, accumulated by the enhanced permeability and retention effect in the tumor. The image time course showcases the benefits of MPI with tracer-like contrast and direct linear quantitation, enabling clear visualization of the particle EPR dynamics with initial rim enhancement, accumulation, and then wash-out. (**b**) Arami et al. 2017 [11] used Lactoferrin functionalized SPIONs injected i.v. at 4 mg/kg for targeting of a C6-rat glioma flank xenograft in mice. After 2 h post-injection, the MPI image showed accumulation in the tumor together with substantial clearance to the liver. Figures reproduced from Arami et al. Nanoscale. 9 with permission from the Royal Society of Chemistry.

The cancer model used seven athymic nude rats bearing flank xenografts of MDA-MB-231-luc breast tumor cells. This work emphasized some of the inherent advantages of MPI for cancer imaging such as excellent image contrast and full quantitation of the tracer dynamics from administration to initial rim enhancement of the tumor, accumulation within the tumor between 1–24 h (peaking at 6 h), and then slow clearance to the liver over a period of 96 h. The signal-to-background ratio of the tumor was very high (>100) as there was no background uptake of SPIONs by biological tissue unlike 18-FDG. Because attenuation correction and signal half-life compensation is not required in MPI, the image quantification was demonstrated to be facile and straightforward. The tracer accumulation in the tumor occurred via enhanced permeability and retention effect (EPR) as this was an untargeted SPION study without cancer-targeting functionalization.

Later in 2017, the first in vivo MPI image of cancer using targeted SPIONs (Figure 4b) was demonstrated by the Stanford School of Medicine Department of Radiology in mice [11]. This study improved targeting to the flank xenograft of C6 brain cancer cells via surface functionalization with lactoferrin and also by placing a permanent magnet on the rodent flank. The SPIONs were multi-modal with Cy5.5 NIRF and 67-Ga radiolabel for near-infrared and SPECT imaging respectively. This study further demonstrated the excellent image contrast of MPI as compared with near-infrared imaging and showed that it approaches the image contrast achievable by 67-Ga SPECT images.

#### *Imaging Cell Therapy for Cancer Immunotherapy Using Magnetic Particle Imaging*

In brief, MPI has many advantages for monitoring of labeled adoptive cell transfer immune cells such as long-lasting magnetic label that does not lose signal over time by radioactive decay and MPI's high-contrast yet quantitative nature. This section introduces cancer immunotherapy and recent efforts to image immunotherapy with MPI.

In the last twenty years, immunotherapy for cancer has steadily gained traction in clinical practice. There are five major types of cancer immunotherapy: (1) cancer vaccines, (2) cytokine therapies, (3) adoptive cell transfer (ACT), (4) immune checkpoint inhibitors, and (5) oncolytic virus therapies [27–31]. Of all these categories, MPI is well-poised to contribute in the adoptive cell transfer category and oncolytic virus category. This is because the magnetic nanoparticles used in MPI have optimal core sizes of 20–30 nm and can thus label immune cells (micron-sized) or oncolytic viruses (150–240 nm). For the adoptive cell transfer category, there is a need to verify that the transferred cells have arrived at and remain at the target cancer site throughout the course of therapy. Furthermore, it is essential to monitor the viability and functionality of the cells to ensure the success of the therapy [32]. These requirements are similar to the imaging requirements for stem cell therapy. Since MPI has been validated in many stem cell therapy studies [33–37], we anticipate MPI's advantages to be applicable to the adoptive cell therapy application as well. The main benefits of MPI in stem cells are innately transferable to the adoptive cell therapy application, such as (1) no loss in signal over time from magnetic cell labels enabling >90% of signal left over 89 days in vivo [33], (2) no radiation dose that will limit the length of a longitudinal study, (3) direct and quantitative measurement of magnetic label that is unaffected by changes in subject anatomy background over time [34], and (4) potential for assessment of viability of labeled cells via color MPI spectroscopic techniques demonstrated in various MPI studies that leverages microenvironment sensitivity for color/contrast change or for multi-contrast multiplexing [38–41]. These initial stem cell studies have demonstrated that the magnetic label remains internalized within the cell population of interest, and that any released label is rapidly cleared to the liver and does not confound the quantitation [33].

ACT has shown the greatest success in "liquid" malignancies such as B lymphocyte leukemia and lymphoma. However, ACT as an immunotherapy for solid tumors has been hampered by an inability to adequately manipulate infused T cells to efficiently traffic into and specifically target deep-seated tumors for destruction, while minimizing immunerelated adverse events (irAEs) caused by low-level recognition of antigen on surrounding healthy tissues [42]. Clinicians thus require real-time information on the biodistribution of ACT products in patients for accurate prognosis and treatment success [43]. MPI of SPION-labeled ACT immune cells can provide high-contrast, sensitive visualization of biodistribution and are thus ideal for this unmet need. The same SPIONs also appear on MRI scans (albeit lower contrast), thus allowing MPI's quantitative nature to complement the high-resolution anatomic MRI scans [44,45]. Rivera-Rodriguez et al. recently demonstrated MPI of ACT immune cells in a mouse model and showed that labeled immune cells showed up in the brain of C57BL/6 mice bearing intracranial KLuc-gp100 tumors 24 h after ACT infusion [46].

Furthermore, ideally immune cells should demonstrate native magnetic signal in order to prevent under-counting that occurs when in vitro magnetic labels are diluted by cell division. Recent efforts tried to overcome this limitation by genetically modifying cells with genes from magnetotactic bacteria [47–49], in order to produce magnetic crystals to enable label-free native magnetic contrast, but this has not been widely implemented on different mammalian cell types yet.

Other than ACT, Magnetic Particle Imaging has also been demonstrated to be helpful in other immunology studies that help advance the field of cancer immunotherapy. For example, the tumor microenvironment is known to greatly impact the success rate and thus a better understanding will help decipher the mechanisms of immunotherapies, define predictive biomarkers, and identify novel therapeutic targets. Figure 5 showcases

recent work on MPI to track tumor-associated macrophages (TAMs). [50,51] Aptly named "Magnetic Particle Imaging of Macrophages Associated with Cancer: Filling the Voids Left by Iron-Based Magnetic Resonance Imaging.", the study showcased how MPI's positive contrast and quantitative nature complements the traditional MRI images of TAMs. In addition to this, MPI can also image inflammation by in situ labeling of inflammatory immune cells [52]. Although this study did not target cancer cells per se, the same in situ labeling concept could be used to image the inflammatory tumor microenvironment.

**Figure 5.** Imaging of Tumor-Associated Macrophages (TAMs) using MRI and MPI. Adapted with permission from Makela et al. Mol Imaging Biol 22. Copyright 2020 Springer Nature. [51]. (**a**) MRI image before non-targeted systemic i.v. injection of 0.5 mmol/kg Ferumoxytol (*n* = 8). L denotes liver, Sp denotes spleen, and T denotes tumor. (**b**) MRI image at 24 h post-injection of Ferumoxytol, where TAMs were seen to have iron uptake due to signal voids visible within the tumor. (**c**) MPI image at 24 h post-injection of Ferumoxytol. The image resolution was lower because Ferumoxytol is not optimal for MPI due to 7-fold worse spatial resolution of the nanoparticle than MPI standard Ferucarbotran. The MPI signal, while visible in the tumor, was not visible in the lung, as the lungs could not be spatially resolved from the liver due to the poor Ferumoxytol spatial resolution. (**d**) Ex vivo MPI of the lungs showed a significantly higher amount of iron in lung metastases compared with healthy control, indicating the presence of iron-labeled TAMs. This study demonstrated that not all MRI iron contrast work well for MPI due to differences in the physical principles of the imaging signal between MRI and MPI.

#### **4. Magnetic-Based Steering and Targeting Strategies Using MPI Hardware**

In brief, MPI use of the strongest magnetic gradients in the imaging field (up to 7 Tesla per meter) equips the MPI scanner to perform magnetic steering of magnetic agents to target tumor sites. This section elaborates on recent efforts to demonstrate this.

To introduce this topic, we must first note that one of the key benefits of using a magnetic imaging agent is the fact that magnetism remains the strongest force-from-a-

distance method for remote steering or targeting [53–55]. There have been many studies of targeting of magnetic entities to a desired in vivo location using strong magnets [56–58]. Both MRI and MPI can benefit from these targeting strategies to enhance the concentration of imaging agent in a region-of-interest for increased binding probability to targets resulting in better imaging or localization of dose for better therapy. For example, Dames et al. 2007 demonstrated the use of a shaped magnetic tip for targeted delivery of magnetic aerosol droplets to the lung [14] (Figure 1e) and Banura et al. 2017 conducted a similar study with the addition of MPI to image the final biodistribution in the lungs after targeting [13] (Figure 6a). Other than the lungs, permanent magnets have been used to enhance delivery to tumors in other parts of the body. Arami et al. 2017 was able to enhance delivery to a flank tumor using an external permanent magnet affixed to the rodent flank [11].

One limitation of these single-magnet strategies is that targeting is only efficient at regions close to the body surface (Figure 1d). Other than embedding a sufficiently strong magnetic dipole source deep within the body, there is no method to magnetically attract towards an arbitrary point in 3D space. However, the hardware of MPI is able to generate a "repulsive" point at an arbitrary point in 3D space. This is because MPI uses a fieldfree-point or field-free-line gradient architecture with rapidly increasing magnetic field strength away from the zero-field-region which implies that magnetic material moves towards the edges of the gradient away from the zero-field-point. Magnetic steering is not unique to MPI, and while MRI has been used to steer large magnetic millimeter-sized ferromagnetic beads in vivo before [59], the weaker gradients used in MRI limit the particle size to about 0.2 mm as smaller objects do not have sufficient magnetic mass for MRI gradients to control [60]. The magnetic force F = Ñ(m·B), where Ñ denotes the change of (m·B) per unit distance with units of m−1. Assuming a magnetically saturated magnetic moment m (constant) as the object with units of Am2, and B as our applied field with base SI units of N A−<sup>1</sup> m−<sup>1</sup> (\*note Tesla = N A−<sup>1</sup> m−1), then in general the magnetic force F scales linearly with the applied field (gradient) strength. MPI's 7T/m gradients [12] are much stronger than the 0.045 T/m gradients used in MRI [2] and can provide much larger magnetic forces for targeting. This has resulted in the capability of MPI scanners to remotely steer catheter tips [61], remotely manipulate an iron screw [62], and in theory also steer particles of sub-micrometer scale. Specifically, magnetic catheter steering has seen clinical usage such as the NIOBE® ES Remote Magnetic Navigation (RMN) System (Stereotaxis, St. Louis, MO, USA) albeit with fluoroscopy imaging. In that clinical application, remote magnetic catheter navigation was performed to guide the catheter through the four heart chambers in order to locally perform atrial fibrillation ablation. Over 200 patients were tested, and it was shown that magnetic steering significantly reduced total fluoroscopy time (10.4 ± 6.4 vs. 16.3 ± 10.9 min) and thus lowered radiation dose to the patient when compared with manual pull-wire catheter navigation [63]. Recent preliminary work in the MPI field has shown some promise to completely replace the fluoroscopy aspect of catheter navigation with non-radioactive magnetic imaging by using one MPI "tracer" to mark the catheter tip and a second MPI "tracer" to replace the iodine contrast that shows the blood vessel size, shape, and branching. By distinguishing the magnetic signatures of the two different "tracers", it enabled interactive magnetic catheter steering with 3D real-time image feedback via "multi-color" MPI [61].

This can be combined with MPI's relatively high temporal resolution of up to 45 fps [64] to enable scan+steer sequences where an image is taken of a volume within 1/45th of a second every second for real-time image feedback of magnetic targeting while dedicating the 44 other frames to holding the magnetically repulsive point in 3D space. With real-time feedback, this can dynamically target the magnetic material towards an arbitrary region in 3D space despite only using a magnetically repulsive point. Proof-of-concept of this simultaneous imaging and MPI-steering of nanoparticles in Figure 6b–d was demonstrated by Griese et al. 2019 in vitro in a bifurcation flow phantom [65].

**Figure 6.** MPI image-guided magnetic steering of magnetic nanoparticles. Figures adapted with permission from Banura et al. Jpn. J. Appl. Phys 2017 [13] under Creative Commons 4.0. (**a**) Magnetic steering of aerosolized magnetic nanoparticles to deposit in a user-selected side of a lung imaging phantom. With closer distance d of the magnetic steering point to the lung, the stronger the accumulation of the magnetic nanoparticles. (**b**) Griese et al. 2020 [65] demonstrated magnetic steering of magnetic nanoparticles in a bifurcation flow phantom. Figures (**b**)–(**d**) adapted with permission Griese et al. J. Magn. Magn. Mater. 498. Copyright 2020 Elsevier. Without the magnetic steering in the control experiment, the dark brown nanoparticle stream bifurcated evenly. Once magnetic steering was turned on, 100% of the particles flowed into the selected right stream. (**c**) One important proof this study demonstrated was that magnetic steering can be performed against a strong flow rate of 6.87 mL/s, as shown by steering of particles into the right arm with a 60% stenosis, although the control clearly shows the flow rate favoring the left arm by order-of-magnitude. (**d**) Seamless switching between steer and image mode was shown in steering particles into an 100% stenosis arm, demonstrating that it is possible to maintain sufficient magnetic force to steer while performing a quick MPI image scan with a time ratio of 20:1 for steer:image.

As testament to the much stronger gradient strengths used in MPI versus MRI, Griese demonstrated that steering against the flow direction is possible by showing steering of nanoparticles into the arm with a 100% stenosis, although the control experiment showed the flow directs the nanoparticles to the non-stenosed arm when the MPI magnetic force is absent. The concept of seamless switching between "steer" and "image" modes was shown too. With a time ratio of 20:1 for force and imaging mode, the induced magnetic force acts for sufficient durations to maneuver the particles towards the stenosis, although no force is acting on the particles during the short time of the imaging mode [65]. In addition, multiple other studies have shown the feasibility of remote magnetic steering of micron-sized objects or synthetic bacteria in vivo [66,67].

Other than using magnetic forces for targeting, some groups have utilized anaerobic magnetotactic bacteria's natural tendency to migrate towards hypoxic regions for targeting hypoxic tumor regions [68]. In this case, the targeting depends on the bacteria, but because the bacteria natively produce magnetic crystals, this can be easily imaged with MPI or MRI. However, other groups have used magnetotactic bacteria for magnetic-field controlled manipulation and actuation of micro-objects [69]. Other strategies do not use the magnetic field for attraction forces but mainly for alignment of travel axis. While the bulk of the

propulsion comes from micro-turbines or flagella [70–72], these micro-swimmers possess a magnetic axis that can be aligned to an external magnetic field for directionality. Unlike MRI having a fixed direction B0 field that limits alignment to the B0 field axis only, MPI's hardware is well-suited here because the "felt" magnetic field lines can be directed in any arbitrary direction by simply shifting the field-free-point gradient field around since the flux lines just surrounding the field-free-point are directed from every direction towards or away from the point.

#### **5. Magnetic Methods for Cancer Therapy in Context of Magnetic Particle Imaging**

Magnetic methods for cancer therapy generally fall into a few categories: (1) Hyperthermia methods that raise the temperature of the cancer cells ranging from mild heating to ablative levels via magnetic nanoparticles, (2) Magnetically actuated drug release from cancer-targeted nanocarriers, or (3) Magnetically actuated mechanical disruption of cancer cells by magnetic particles or magnetic micro-/nano-robots. In this section, each category is discussed and the benefits and relevance of Magnetic Particle Imaging towards these methods is explained.

#### *5.1. Magnetic Hyperthermia Therapy (MHT)*

In brief, MPI unique scanner architecture gives it the potential to be integrated with the alternating magnetic field (AMF) coils used for magnetic hyperthermia, enabling seamless image-guided therapy workflows. Another unique point is that the pre-existing gradients on the MPI scanner can be used to focus magnetic hyperthermia solely at the field-freepoint (FFP) or field-free-line (FFL), enabling unprecedented targetable precision at-depth and in a 3D manner. This section explains the background of hyperthermia and recent MPI efforts to synergize with MHT.

The general principle of hyperthermia is based on increasing the temperature of a tissue of interest above 40 degrees Celsius [73,74]. While there are several methods to increase the temperature in hyperthermia, including microwaves, ultrasound, and laser, we focused on radiofrequency magnetic hyperthermia in this article. Magnetic hyperthermia (MH) is a promising cancer therapy that is induced by applying an alternating magnetic field (AMF) of frequencies ranging between 100 kHz and 1 MHz into magnetic nanoparticles targeted in the tumor area [75]. Under such conditions, magnetic nanoparticles act as very local heat sources, which are capable of raising the temperature of cancer tissues and consequently destroying the tumor in a localized and effective way. The heat generated by both superparamagnetic or ferromagnetic nanoparticles is originated from hysteresis losses and is proportional to the area of the hysteresis loop described by the magnetic nanoparticles during the application of the AMF [76,77].

The key advantages of MH are (1) the ability to treat at deeper regions of the body where other surface methods like microwaves, ultrasound, and radiation cannot, (2) negligible energy dose is deposited in healthy tissue en route to the target site as almost all the heat dose comes from the magnetic material on-site, (3) the magnetic material is not consumed by the therapy and allows for multiple treatment sessions per injection, and (4) the thermal dose is externally controlled by the AMF applicator which can compensate for variability in magnetic material accumulation at cancer site to ensure correct thermal dosing [78].

The first application of Magnetic Hyperthermia was in 1957 in dogs, where it aimed to treat cancers that had metastasized to the lymph nodes [79,80]. Most of the subsequent studies relied on direct injection of magnetic material into the tumor [81,82] rather than systemic delivery. To address this issue, Ivkov et al., in 2005, utilized monoclonal antibody targeting to cancer tissue [83]. Various other groups used magnetic nanoparticles within cationic liposomes for efficient accumulation into tumors and demonstrated therapeutic effect in rat glioma [84–86], melanoma [87,88], and prostate [89] animal tumor models. In recent years, an increasing number of in vivo and in vitro works have been reported in the literature [21,90]. In 2001, Jordan et al. showed the treatment of human solid tumors with MFH [91]. Due to the obtained promising results, several clinical trials have been carried out for the treatment of glioblastoma multiforme and prostate cancer. In 2003, the first phase I clinical trial was performed on 14 patients with glioblastoma multiforme (GBM) at the Charité Hospital in Berlin (MagForce Nanotechnologies) [77,91–93]. In 2005, Johannsen et al. reposted the first phase I clinical trial carried out in 10 patients with locally recurrent prostate cancer [94–96]. In 2010, MagForce AG obtained European Union Regulatory Approval (10/2011) for its the Nanotherm® therapy and later in 2013 started a clinical study in current gliobastoma with Nanotherm® therapy after receiving approval from the German Federal Institute for Drugs and Medical Devices. Recently, the FDA approved a single-arm study of NanoTherm (R) therapy system for intermediate-risk prostate cancer [97].

Despite all these clinical trials, there are several challenges that need to be addressed. One issue of MH is related to low accumulation of magnetic nanoparticles at the tumor site [98]. In order to achieve an efficient magnetic hyperthermia treatment, the heating efficiency (also known as Specific Absortion Rate (SAR)) of magnetic nanoparticles needs to be as high as possible in order to destroy the cancer with the low amount of magnetic nanoparticle available in the target site. The SAR greatly depends on the physicochemical properties of the nanoparticles such as composition, size, shape, crystallinity, and saturation magnetization [99,100]. Additionally, interparticle magnetic interactions, the interplay between particles and biological systems, and AMF parameters also affect the heating performance of magnetic nanoparticles [101,102].

Currently, different approaches have been proposed in the literature to design magnetic nanoparticles that exhibit high SAR values. Tailoring the shape of the magnetic nanoparticles can provide an effective strategy to increase their heating efficiency. For instance, Guardia et al. showed that the 19 nm iron oxide nanocubes possess very high SAR values (up to 2452 W/g at 29,000 A/m and 520 kHz) compared with spherical particles of similar size [100]. Other promising designs include magnetic vortex nanorings reaching 3000 W/g (at 64,000 A/m and 400 kHz) with demonstrated efficacy in vivo [103]. Some studies also use exchange-coupling between a magnetically hard core and magnetically soft shell to enhance SAR values (3886 W/g at 37,000 A/m and 500 kHz) to an order-ofmagnitude greater than conventional iron-oxide nanoparticles, with superior therapeutic effectiveness in mice tumor models over chemotherapeutic drugs [104]. In addition, tuning the arrangements formed by dipolar interactions can also help enhance the heating efficiency of magnetic nanoparticles. Some works in literature have reported that specific arrangements formed by dipolar interaction, like chain-like structures, increase the SAR due to their ability to mechanically orient along the field lines [105,106]. Gandia et al. [107] proved that magnetotactic bacteria of the species M. gryphiswaldense, which internally biomineralized magnetosome chains, give rise to very high SAR values, up to 2400 W/g at 28,000 A/m and 300 kHz. Table 1 provides a summary and key characteristics of these MHT agents.

**Table 1.** Summary of studies demonstrating MHT agents with high SAR values.


Magnetic Particle Imaging provides key benefits for MHT such as image-guidance (Figure 1f), quantitation of magnetic material on-site, which is essential for MHT thermal dose planning, and also the ability to select which magnetic nanoparticles to heat with pinpoint precision as low as a few millimeters [15,108–110]. This precision capability is a novel benefit in the field of MHT for cancer therapy. To explain further, consider chemotherapy and radiotherapy, which benefit from a significant differential cytotoxicity between cancerous and healthy cells [111]. Despite this, significant side-effects still exist due to collateral damage to healthy tissue. A similar issue exists for MHT where nanoparticle targeting/trafficking to tumors is not perfect and healthy tissue also accumulate nanoparticles. The additional precision in magnetic excitation enabled by MPI thus greatly mitigates collateral thermal damage to off-target healthy tissues frequently caused by magnetic particles biodistributed to other sites in the body, especially clearance organs such as the liver or spleen. This indirectly increases the therapeutic ratio to allow higher nanoparticle dosage as the side-effects to healthy cells are minimized. This concept is also seen for targeted nanocarriers for drug delivery, where precision of drug release enables higher doses while having lesser side-effects. Details are shown in the next section of this article. Conventional external AMF applicators used in MHT are unable to target magnetic excitation and heating only to the tumor because the long wavelength of the AMF at about ~50 m precludes the possibility of lens-based focusing of the magnetic field at a distance [15]. Other attempts using an array-based synthesization technique were able to project a focal point AMF at a distance of 10 cm but precision remained low with spot sizes of 5 cm [112,113]. Improving the precision to 2.5 cm required exponentially high currents in the kilo-ampere range [113]. In contrast, the mechanism for MPI's precision heating relies not on "focusing" the AMF into a narrow spot, but rather it suppresses the heating capability of off-target magnetic material by magnetically saturating off-target material so that it cannot respond and get heated by the AMF [15,114–116]. This can be achieved by MPI's field-free-line or field-free-point gradient hardware where the precision linearly scales with the gradient strength [115,116]. For example, in Figure 7b, at a gradient strength of 2.35 T/m, precision of 7 mm was achieved [15]. The field-free-region (zero field point) was simply placed over the target spot, enabling only that point-in-space to respond to AMF while magnetically saturating all other regions in space. Because the hardware for precision targeting exists within the MPI scanner and because the MPI scan at 20 kHz is demonstrated to have zero heating of particles [15], MPI is innately suited for image-guided precision MHT by simply imaging at 20 kHz then switching to a ~300 kHz for gradient-targeted precision MHT. Considering MPI's fully quantitative imaging of magnetic nanoparticle mass, it is possible to develop the ideal MHT workflow of (1) image, (2) quantitate, (3) dose planning, (4) target positioning, and (5) precision MHT all within a single MPI scanner. This ideal workflow was demonstrated in a rodent cancer model by Tay et al. 2018 (Figure 7a), where the efficacy of precision MHT and mitigation of collateral thermal damage to the liver was validated in vivo [15] (Figure 7c–e).

**Figure 7.** MPI image-guided precision Magnetic Hyperthermia with non-invasive pinpoint heating to 2–7 mm precision at a depth of 4 cm. Adapted with permission from Tay et al. 2018 [15]. Copyright 2018 American Chemical Society. (**a**) Theranostic workflow with MPI image guidance, selection of tumor target, and pinpoint localized heating. (**b**) Precision of at least 7 mm was demonstrated where any user-selected well in a custom 9-well plate can be heated to a 30 degrees increase in temperature with negligible heating in all adjacent targets 7 mm away. (**c**) In vivo results show user-selected precision heating of one of two adjacent tumors to the exclusion of the other. (**d**) Bioluminescence as a viability marker of luc-competent tumor confirms pinpoint therapy of one tumor with negligible impact on off-target tumor. (**e**) Although the mouse liver received a significant nanoparticle dose as shown in a, apoptosis assay showed that MPI hyperthermia (third column) improved precision-to-tumor over conventional hyperthermia (fourth column) which, due to its wide area magnetic excitation, collaterally damaged the liver while treating the tumor. This precision capability can reduce the side-effects of damage to healthy tissue in clinical settings.

#### *5.2. Magnetically Actuated Drug Release*

In brief, MPI can provide image-tracking of the magnetic-labeled drug delivery platform in vivo, ensuring that arrival at the target tumor site has occurred before triggering drug release by magneto-mechanical or magnetic heating in the case of thermosensitive liposomes. Similar to Section 5.1, MPI's selection field (FFP or FFL) can localize the triggering to only the field-free-region, increasing precision of therapy and further reducing drug side-effects. Finally, MPI images during therapy provide real-time feedback on the extent of drug released from the carrier. This section reviews recent efforts of MPI in the drug delivery field.

Chemotherapy has been one of the mainstays of cancer therapy and there has been much work in developing targeted nanocarriers with controlled release of chemotherapeutic drugs at the tumor to reduce systemic toxicity while maximizing the drug dosage at close proximity to the tumor to improve the therapeutic index [117,118]. Several methods to actuate the release of the chemotherapeutic have been developed and can be widely classified into external stimuli (magnetic, ultrasound, electric field, thermosensitive, UV–vis light, etc.) or endogeneous stimuli (pH-sensitive release, cancer-linked enzyme cleavage

reactions, redox reactions, etc.) [119]. Magnetic methods to actuate drug release have several benefits over other methods such as (1) the ability to access deeper regions of the body with no view limitations and (2) the relative safety of magnetic fields compared with other methods for external stimuli that may affect healthy tissue en route to the target [120]. There have been many studies detailed below showing the efficacy of magnetic actuation for controlled chemotherapeutic release. The mechanism relies on a magnetic force to mechanical energy conversion and in many cases there is no detectable temperature rise, although it is also possible to combine both mechanical and MHT heating to doubly trigger release. In 2012, Peiris et al. developed a multi-component iron oxide nanochain with radiofrequency-tunable drug release [121]. The magnetic nanochain efficiently converts magnetic energy from a 10 kHz, 1–50 W external magnetic field into mechanical vibrations that trigger drug release from the attached DOX-loaded liposome. The release rate could be modulated by the operating parameters of the magnetic field. A temperature-sensitive fluorophore attached to the chain acted as a thermometer to verify the absence of local heating. In 2013, Oliveira et al. showed magnetic field triggered drug release (14 mT 750 kHz) from polymersomes, which are notable for their ability to load both hydrophilic and hydrophobic drugs [122]. In 2018, Nardoni et al. used pulsed magnetic fields (20 kHz, 60 A/m) to actuate drug release from high transition-temperature (Tm = 52 ◦C) magnetoliposomes [123]. The transient increase in membrane permeability upon actuation allowed on-demand drug release while ensuring negligible leakage and safety at all other times. For magneto-thermal mechanisms of drug release, Fuller et al. 2019 demonstrated nanocarriers with a hydrophobic core of superparamagnetic iron oxide nanoparticles that released heat upon AMF to actuate release of drug cargo from a thermoresponsive polymer based on thermally labile Diels-Alder bonds [124].

Magnetic Particle Imaging provides several key benefits for magnetic drug release— (1) Image-guidance and quantitative assessment of nanocarrier accumulation at target tumor site, (2) pinpoint precision of a few millimeters in actuating drug release while suppressing drug release from off-target nanocarriers (Figure 1g), and (3) real-time feedback on the amount of drug released from the magnetic nanocarrier via changes in the magnetic component's MPI spectrum. Similar to MHT, (1) is crucial for dose planning, especially when the amount of drug release is tunable such as in the study led by Peiris [121] (Figure 8a,b). Benefit (2) works on similar principles to that earlier described for MHT, where the suppression of off-target magnetic entities via magnetic saturation also works to suppress the induced mechanical forces. In other words, the magnetic components are overwhelmed by the background gradient magnetic field and are thus aligned and locked to the directionality of the background magnetic field. MPI's most unique benefit can be considered to be the real-time feedback on the amount of drug release. The mechanism of this depends on the different microenvironment around the nanoparticles within the nanocarrier as opposed to free nanoparticles after rupture of the nanocarrier. The particles report the change in microenvironment by a quantitative shift in the MPI spectrum. There has been much work in the MPI field to make these "color MPI" algorithms robust and quantitative to microenvironment factors, i.e., viscosity, pH, and inter-molecular binding [38–41].

Combining benefit (3) with the ability to switch between imaging 20 kHz and actuation ~300 kHz on the same MPI scanner, it is possible to develop the ideal therapy workflow within a single scanner.

The ideal workflow of (1) image, (2) quantitate, (3) dose planning, (4) target positioning, (5) precision drug release, and (6) real-time feedback (Figure 1h) on the amount of drug released would be desirable for controlled drug release applications. This workflow is theoretically feasible, although no one group has demonstrated the entirety of this workflow. Separate groups have proof-of-concept studies on each step of the workflow. Maruyama et al. 2016 demonstrated MPI quantitation of magnetic nanocarriers based on a thermoresponsive liposome design [125]. Liu et al. 2018 demonstrated target positioning and precision drug release (Figure 8c,d) at millimeter-scale precision from magnetic nanoliposomes with MPI-like gradient fields (referred to as static gating fields in this paper) [16]. Zhu et al. 2019 used MPI for in vivo quantitative drug release monitoring in tumors of a murine breast cancer model to measure in real-time the amount of drug release [17] (Figure 9a–c). Finally, MPI can be used to monitor apoptosis in tumors post-treatment. Using an apoptosis-specific tracer, MPI can accurately quantify apoptosis as the imaging signal was almost proportional to the number of apoptotic cells [126] (Figure 9d).

**Figure 8.** MPI for magnetically-actuated drug release [121]. Top row figures adapted with permission from Peiris et al. ACS Nano (2012) [121]. Copyright 2012 American Chemical Society (**a**) Liposome attached to a nanochain of three SPIONs that mechanically oscillates when exposed to an alternating magnetic field. Rather than using thermo-sensitive release, this work shows the feasibility of mechanical energy for rupturing the attached liposome. (**b**) The extent of drug release can be finely-tuned and controlled by the AMF frequency and power. Bottom row figures adapted with permission from Liu et al. Small (2018) [16]. Copyright 2018 WILEY-VCH Verlag GmbH. (**c**) Magnetic gradients (also termed static gating field in this article) can be used to target drug release to selected locations with 2 mm precision while suppressing release from other neighboring nanocarriers. Since MPI have the strongest magnetic gradients in imaging, the same concept demonstrated in Figure 7 can be replicated here for image-guided targeting of drug release. (**d**) Fluorescence imaging of released DOX from thermosensitive liposomes verifies that only the targeted well triggered drug release.

**Figure 9.** MPI for monitoring of the percentage of drug release from nanocarrier in vivo. Adapted with permission from Zhu et al. Nano Lett (2019) [17]. Copyright 2019 American Chemical Society. MPI enables non-invasive imaging assessment of the extent of drug release via MPI signal differences when SPIONs are encapsulated within a pH-sensitive nanocarrier and after their release together with drug upon nanocarrier rupture. (**a**) In vivo MPI images of nanoparticle-PLGA-Doxorubicin nanocarriers in vivo showing increasing MPI signal over a span of 48 h. (**b**,**c**) Quantification of MPI image intensity and percentage of DOX release from the nanocarrier shows a good correlation between MPI signal intensity and the percentage of release, verifying that the designed nanocarrier works as intended to have low MPI signal pre-release and high MPI signal after the nanoparticles are freed from the nanocarrier together with the Doxorubicin. (**d**) MPI of AnnexinV-SPION that binds to apoptotic cells in mouse xenograft model post-therapy, showing that MPI can evaluate the anti-tumor efficacy of cancer therapy. Figure (**d**) adapted with permission from Liang et al. Phys Med Biol. 2020. [126]. Copyright 2020 by Institute of Physics and Engineering in Medicine. Reproduced by permission of IOP Publishing. All rights reserved.

#### *5.3. Magnetically Actuated Mechanical Disruption of Cancer Cells*

In brief, other than magnetic hyperthermia and actuation of drug release, MPI can provide the magnetic energy and control necessary to actuate mechanical disruption of cancer cells. MPI's unique scanner architecture is well-suited for this because it already has 3-axis drive coils capable of up to 25 kHz and 25 mT field strength to power the magnetic actuation in various directions or to produce rotating magnetic fields. Most importantly, unlike MRI, which has an "always-on" main field forcing a fixed alignment of all magnetic material axes within the MRI scanner, MPI can turn off the electromagnet selection field and allow the drive fields to establish magnetic control of mechanical agents in vivo. This section reviews magneto-mechanical methods for cancer and explains how MPI's electromagnets can specifically achieve magnetic actuation for these methods.

To introduce the magneto-mechanical approach, we must first note that the magnetic forces incident on magnetic particles can be translated into mechanical energy that directly destroys cancer cells. Creixell et al. 2011 demonstrated that EGFR-targeted magnetic nanoparticles under AMF excitation were able to kill cancer cells (at a 99.9% loss in viability) without a perceptible temperature rise [127]. This runs contrary to the expectation that a temperature rise of up to 43–46 ◦C is needed to kill the cells under AMF. Because binding and subsequent activation of EGFR is implicated in cancer cell apoptosis, the EGFnanoparticles without application of AMF already demonstrate some toxicity to the cancer cells. However, the application of AMF significantly increased the toxicity and suggests that magneto-mechanical stimulation of the EGFR via the attached EGF-nanoparticle greatly upregulates the relevant apoptotic pathways. This showcases the feasibility of magnetomechanical actuation of apoptotic pathways in cancer. Other than EGFR, overactivation of ERK proteins via magnetic particles was also investigated to stop the cancer cell cycle of replication.

Besides mechanical activation of receptor-linked pathways, a more direct method is the mechanical disruption of cellular structures. Externally-bound magnetic particles can compromise cell membrane integrity promoting cell lysis, while internalized magnetic particles cause perturbations in lysozymes leading to enzyme-based cell suicide or damage the cytoskeletal integrity of the cell [128]. Liu et al. 2012 used magnetic carbon nanotubes at 75 mT 16 Hz for magnetoporation of the cancer cell membrane, as measured by increased membrane roughness by Atomic Force Microscopy and Scanning Electron Microscopy [129]. Wong et al. reported similar membrane integrity alteration with magnetic NiFe nanowires at 14 mT 5 Hz via ethidium bromide staining [130]. Domenech et al. showed lysosomal membrane permeabilization in cells that internalized iron oxide magnetic particles with increased the release of proteolytic cathepsin B activity leading to the cancer cell to selfdigest [131]. Zhang et al. and Shen et al. reported similar results [132,133], but Master et al. reported negative results where lysosomal disturbance was not observed [134]. Master et al. also targeted the cell cytoskeletal component actin, harnessing the observation that cancer cells are less rigid than healthy cells. The results showed cancer cells were more susceptible to cytoskeletal disruption by actin-targeted magnetic particles under AMF [134]. Additionally, disk-shaped magnetic particles have also been used to magnetomechanical damage cancer cell integrity [135]. The disks can be actuated by an external magnetic field to exercise mechanical force on the cancer cell. Kim et al. used Ni80Fe20 microdisks with magnetic vortex configuration for in vitro experiments with glioma cancer cells. In this work, a 90% of cell death was reported after applying 9 mT and 10–20 Hz during 10 min [136]. Goriena et al. used Ni80Fe20 vortex configuration nanodisks [137] almost ten times smaller than those used by Kim et al. to destroy lung cancer cells [135]. The application of a 10 Hz oscillating magnetic field of 10 mT during 30 min reduced the cell by 30%. Beside disk-shaped magnetic particles with vortex state, perpendicularly magnetized synthetic antiferromagnetic (P-SAF) disks have also been used for cancer treatment through mechanical cell disruption [138].

Magnetic actuation was also demonstrated to be useful in a more macroscopic scale. For example, magnetic microbots at micron-level sizes [139–141]. Lee et al. 2020 demonstrated a micron-sized nickel-based magnetic corkscrew that is actuated by an external rotating magnetic field to "corkscrew" itself onto the cancer cell [142]. The microrobot then releases chemotherapeutic drugs after affixing itself to the target cell. Vyskocil et al. 2020 developed Au/Ag/Ni microrobotic scalpels that enter and exit an individual cancer cell and cut the cancer cell under actuation by an external rotating magnetic field [143]. This is relevant to MPI scanners because these rotating magnetic fields can be achieved with the 3-axis electromagnets used to produce the Lissajous trajectory for the FFP. Betal et al. 2018 developed a core-shell magnetoelectric nanorobot that uses DC magnetic gradients for navigation and steering to the target cell. This is relevant to MPI scanners because the

DC magnetic gradients needed can be fulfilled by the MPI selection field gradients. The same nanorobot under AMF actuation transforms into a localized electric-pulse generator for targeted cell electroporation. This can directly kill the cancer cells or increase their susceptibility to chemotherapeutics. [144].

As mentioned in previous chapters, MPI systems can provide image-guidance for this type of cancer strategy while offering much stronger AMF (25 mT at 20 kHz) and background gradient capabilities (up to 7 T/m) than MRI systems that typically have a 0.045 T/m gradient and microtesla RF excitation as AMF. While MRI systems can be enhanced with in-bore additions such as the shaped soft-iron core used in dipole-field navigation [145], the tradeoff in image quality due to susceptibility artifacts is usually substantial [60]. As such, MPI's unique hardware makeup as an imaging modality utilizing strong AMF and strong gradients enables its theranostic capabilities too, leading to better magnetic targeting, control, and actuation for magneto-mechanical strategies to kill cancer cells.

#### **6. Safety of MPI and Current Status of Clinical Translation**

MPI has comparable safety to MRI, which has been widely recognized as a safe medical imaging modality because it utilizes safe magnetic fields for excitation and signaling. Magnetic fields are non-ionizing and pass through the human body safely without attenuation or any mechanical tissue destruction. There are only two main safety considerations: (1) magnetic stimulation of peripheral nerves causing tingling sensations at the body peripheries when the alternating magnetic field strength is too high and (2) eddy-current induced warming of tissue when the magnetic field is at radio-frequency operating range, which is also known as SAR-related safety limits. For MPI, the drive field operating frequency is relatively low between 1–50 kHz and therefore close to the 42 kHz junction of magnetic stimulation (magstim) safety limits and SAR limit dominance as outlined by the safety study on human volunteers performed by Saritas et al. [146]. This MPI tailored safety study by Saritas et al. is most suitable for MPI's 20 kHz drive fields that finds no precedence in MRI safety standards. The results showed a limit of 15 mT peak-to-peak for the drive field, which is amenable to MPI imaging-parameters and therefore there are no fundamental safety concerns for MPI.

We can also evaluate MPI safety from the viewpoint of international commission on non-ionizing radiation protection (ICNIRP) standards. For instance, a maximal value of 2.7 Tesla per second for dB/dt (1 Hz–3 kHz, applicable to gradient fields MRI/MPI) was recommended to prevent any magnetostimulation or magnetophosphor effects on patients [147]. MPI's selection gradient fields, which are shifted by mechanical motion of the patient or by electromagnets at the range of 1 Hz to 3 kHz, stay within the limits proposed. Even for a very strong 7 Tesla per meter selection gradient field, the shift rate can be as high as 0.38 m per second, sufficient to raster the FFP or FFL across a typical clinical FOV dimension of 38 cm in 1 s. For a weaker selection field, this can be proportionally faster. Thus, MPI's gradient fields have no safety issues under the ICNIRP standards applied to MRI gradient fields.

For Magnetic Hyperthermia, the limits are dominated by the SAR-related safety regime and the safety limits are well-defined by general Atkinson-Brezovich limit (H × f <= 5 × <sup>10</sup><sup>8</sup> Am−<sup>1</sup> <sup>s</sup><sup>−</sup>1) [148] and the Hergt criterion [149], which is less rigid and tailored to the area of application of the body (H × f <= 5 × 109 Am−<sup>1</sup> <sup>s</sup>−1). To give examples of existing devices in Johns Hopkins University, Attaluri et al. 2020 constructed a Maxwell-type induction coil prototype for magnetic nanoparticle hyperthermia in phantoms and large animals. The prototype was designed to be scalable to a human-sized system (60 cm diameter) [150]. For hyperthermia devices in Berlin, MagForce AG obtained European Union Regulatory Approval (10/2011) for its Nanotherm® therapy, and clinical studies for glioblastoma therapy were performed in 2013 with this approved instrument. More recently, MagForce received 2020 FDA approval for use in intermediate-risk prostate

cancer. These examples and associated references were previously discussed in detail in Section 5.1.

The iron oxide nanoparticles used in MPI have a long history of safe usage in medicine, both as MRI tracer agents and for treating anemia. Iron oxide nanoparticle is considered safe and specific anaphylactic reaction observed when used is often associated with the parenteral formulation (can be made safer) and not the magnetic core that produces the MPI signal [151]. The nanoparticles used in MPI do not contain toxic magnetic elements like cobalt and are entirely iron-oxide based for biocompatibility. Some examples of clinically approved iron oxide are Ferumoxytol (USA) and Ferucarbotran/Resovist® (Japan) and the latter has been shown to work well for MPI [37,152,153]. Regarding clearance, iron oxide nanoparticles are easily assimilated by the liver and spleen and cleared by the hepatobiliary system [154]. Any digested iron from the particles becomes incorporated in the porphyrin rings of hemoglobin [151], replenishing the blood with iron-rich hemoglobin and forming the basis for treating anemia [155].

Although the MPI field has mostly shown preclinical studies, recent work has shown that the imaging technology can be scaled-up to clinical scale. Graeser et al. 2019 showcased a human-sized MPI scanner for brain imaging applications [156]. Mason et al. 2020 showcased an MPI design for clinical intraoperative applications [157].

#### **7. Conclusions**

Magnetic Particle Imaging is an emerging imaging modality with numerous complementary aspects to the more established MRI in the field of magnetic methods for cancer. Other than direct imaging of tumors, MPI has shown promise to value-add to passive nanocarriers [22] in other aspects such as targeting enhancement, actuating therapy, and post-therapy monitoring. Existing magnetic nanoparticles have mostly been optimized for MRI imaging, but with recent work on optimizing nanoparticles for MPI and heating theranostics, we hope that the capabilities of MPI can be significantly enhanced in the near future by these new classes of nanoparticles. The MPI engineering field has also shown great progress towards clinical translation with recent work showcasing a human head-sized MPI scanner. Overall, MPI has demonstrated its potential in a wide range of applications from tumor imaging to magnetically-actuated in situ drug release. With good compatibility for immunotherapy cell labeling, intrinsic high gradient strengths for magnetic steering and targeting, and finally the capability for spatially precise AMF magnetic heating/actuation, MPI shows great promise as a magnetic platform technology for cancer theranostics.

**Author Contributions:** Conceptualization, Z.W.T. and S.M.C.; writing—original draft preparation, Z.W.T.; writing—review and editing, Z.W.T., P.C., B.D.F., I.R.A., E.Y., M.O., S.M.C.; supervision, M.O., S.M.C.; funding acquisition, M.O, S.M.C. All authors have read and agreed to the published version of the manuscript.

**Funding:** For affiliation #2, we gratefully acknowledge funding support from National Institutes of Health, NIH1R01EB024578-03, 3R01EB024578-03S1, R44EB029877-01A1 and 1R24MH106053-01. For affiliation #1, we acknowledge funding support from Agency for Science, Technology and Research's (A\*STAR) BMRC Central Research Fund (CRF, UIBR) Award and CDA grant 202D800036.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

