Magnetic Hyperthermia for Cancer Treatment: Main Parameters Affecting the Outcome of In Vitro and In Vivo Studies

Magnetic hyperthermia (MHT) is being investigated as a cancer treatment since the 1950s. Recent advancements in the field of nanotechnology have resulted in a notable increase in the number of MHT studies. Most of these studies explore MHT as a stand-alone treatment or as an adjuvant therapy in a preclinical context. However, despite all the scientific effort, only a minority of the MHT-devoted nanomaterials and approaches made it to clinical context. The outcome of an MHT experiment is largely influenced by a number of variables that should be considered when setting up new MHT studies. This review highlights and discusses the main parameters affecting the outcome of preclinical MHT, aiming to provide adequate assistance in the design of new, more efficient MHT studies.


Introduction to Magnetic Hyperthermia: Concepts and Terminology
The term "hyperthermia" refers to the local, regional, or generalized increase in body temperature, and it has a solid, long-lasting history in the annals of cancer management, either alone or in combination with other therapeutic approaches.
The use of magnetic implants as thermo-seeds to generate heat when exposed to an alternating magnetic field (AMF) has been proposed as cancer treatment since the 1950s [1]. The concept of "intracellular" magnetic hyperthermia (MHT) was later introduced by Gordon et al. by using dextran-coated magnetite submicron particles, which were internalized by cancer cells in vivo, to increase the temperature of tumors submitted to a strong AMF [2]. More recently, nanotechnology has been significantly contributing to the on-going scientific progress in the field of cancer research, not only regarding treatment but also its prevention and detection [3][4][5]. Innovative approaches have been proposed to treat cancer, among them magnetic nanoparticle (MNP)-induced MHT, which explores the heating ability of MNPs, under the influence of an AMF, to kill cancer cells (Figure 1). Some of the expected theoretical advantages of MHT are the possibility of treating a localized area, while keeping surrounding tissues safe, the ability to treat deep-seated tumors that would otherwise be untreatable and the possibility of exploring combinatorial schemes with other therapeutic regimes for increased efficiency [6].  General schematic representation of a magnetic-hyperthermia study (a) in vitro and (b) in vivo.
Regardless of the clinical interest and the large availability of scientific literature on the topic, MHT for clinical treatment is hardly a reality and the number of preclinical studies progressing to clinical trials is minimal. Establishing comparisons between MHT studies is a challenge due to the widely different conditions used by different authors, which, in some cases, is aggravated by the lack of crucial information concerning a certain aspect of the procedure. In this review, the main parameters influencing the outcome of preclinical MHT studies will be addressed and discussed with the purpose of providing a source of helpful information for planning forthcoming MHT studies.

Main Parameters Influencing the Outcome of a Preclinical Magnetic Hyperthermia Study
The approval of nanomaterials for cancer clinical trials requires initial testing of their anti-cancer potential in vitro and in vivo in order to have an initial impression on the efficacy, tolerability, and toxicity of the MNPs and the selected AMF conditions. In general terms, an MHT experiment comprises three distinct layers: a biological component (cellular and/or animal models), a nanomaterial component (most commonly, MNPs), and an AMF component (considering the MHT equipment and the derived AMF specifications). A fourth level concerns the evaluation of the treatment outcome.
Some parameters will crucially impact the observed results, justifying the observed colossal differences in the outcome of MHT studies. Among those parameters, we highlight the cells to be tested, the MNPs' characteristics (namely, the coating employed, the size, and initial concentration), the targeting system (if applicable), the selected animal model (if applicable), the AMF parameters, and the temperature that was reached. The evaluation of the MHT effect, i.e., which kind of test to perform, and the most appropriate time to implement it, also influence the reported outcomes. Tables A1 and A2 compile examples of in vitro MHT studies, using non-targeted and targeted nano-formulations, respectively, while Table A3 concerns in vivo studies. The most relevant parameters that may have contributed to the observed outcome are summarized and will be further examined and discussed in this review.

Relevance and Thermal Susceptibility
Since MHT is often purposed for cancer treatment, it is important to test its tolerability and its efficiency to kill cancer cells that are representative of a tumor, both in vitro and in vivo. These are usually human cell lines that need to be tumorigenic to generate a tumor in an animal model. For some types of cancer, e.g., for glioblastoma, there are a number of different human cell lines that fill in this pre-requisite. However, the results of MHT studies often lead to distinct outcomes. In fact, different cells or cell lines may have diverse susceptibilities to heat, i.e., thermotolerance [7], which means that similar temperature profiles may lead to varying outcomes depending on the tested cells [8]. One major contributing factor for such differences may be the divergent induction of heat shock protein (HSP) synthesis among cell lines [9]. Some authors are actively trying to overcome this issue by blocking HSP70 to enhance MHT efficiency [10].

Cell Number, Configuration, and Tumor Size
Since cancer cells tend to grow fast, the initial number of cells used for the experiments will have an impact on the observed results. An increased number of cells is able to uptake a larger total amount of magnetic material and, consequently, generate more heat when subjected to AMF. This means that using different number of cells might result in a different concentration of magnetic materials with all the associated consequences. This further applies to an in vivo scenario, as the number of tumorigenic cells inoculated in the animal model will result in variable tumor sizes.
The configuration in which the cells are exposed to MNPs and subjected to MHT may also result in different outcomes. In fact, under similar thermal loading, adherent cells seem to be more susceptible to MHT than suspended cells [11]. While in a cell suspension, the heating effects may be mediated by heating the surrounding medium, the real-time follow-up of the treatment on adherent cells revealed the delayed-onset of cell death occurring on a cell-by-cell basis, which, therefore, supported an intracellular type of MHT.
Other conformations of cells have been used for MHT studies like cell pellets [12] and cell clusters [13]. These cell clusters helped understand the influence of the tumor size on the efficiency of MHT, concluding that a minimum tumor volume of 1 mm 3 is required for cytotoxic hyperthermia [13]. This result suggests the inability of MHT to treat microscopic tumors or metastasis.
In an attempt to better emulate the complex 3D structure of a tumor in vitro, Stocke and co-workers developed a co-cultured spheroid system containing triple negative breast cancer cells, endothelial cells, and embryonic fibroblasts from murine origin [14], which recreated a secondary lung tumor of a metastatic breast malignancy. After 10 days in culture (600 µm in diameter), these 3D structures were exposed to a low or a high concentration of inhalable MNPs and to AMF for 1 h, which resulted in a dose-dependent increase in cell death. The reached temperature was not reported and the contribution of heat for the observed outcome was not excluded, but the authors suggest it to derive from mechanical deterioration. Very recently, Mamica et al. proposed a new in vitro model consisting of a tumor-on-a-chip to study the effects of MHT on treating glioblastoma ( Figure 2) [15]. The tumor was mimicked by using rat glioblastoma cells in 3D configuration in a device integrating microfluidics that guarantees the delivery of nutrients to the tumor. Future developments would include the emulation of the blood vessels in this model.

Cell Number, Configuration, and Tumor Size
Since cancer cells tend to grow fast, the initial number of cells used for the experiments will have an impact on the observed results. An increased number of cells is able to uptake a larger total amount of magnetic material and, consequently, generate more heat when subjected to AMF. This means that using different number of cells might result in a different concentration of magnetic materials with all the associated consequences. This further applies to an in vivo scenario, as the number of tumorigenic cells inoculated in the animal model will result in variable tumor sizes.
The configuration in which the cells are exposed to MNPs and subjected to MHT may also result in different outcomes. In fact, under similar thermal loading, adherent cells seem to be more susceptible to MHT than suspended cells [11]. While in a cell suspension, the heating effects may be mediated by heating the surrounding medium, the real-time follow-up of the treatment on adherent cells revealed the delayed-onset of cell death occurring on a cell-by-cell basis, which, therefore, supported an intracellular type of MHT.
Other conformations of cells have been used for MHT studies like cell pellets [12] and cell clusters [13]. These cell clusters helped understand the influence of the tumor size on the efficiency of MHT, concluding that a minimum tumor volume of 1 mm 3 is required for cytotoxic hyperthermia [13]. This result suggests the inability of MHT to treat microscopic tumors or metastasis.
In an attempt to better emulate the complex 3D structure of a tumor in vitro, Stocke and coworkers developed a co-cultured spheroid system containing triple negative breast cancer cells, endothelial cells, and embryonic fibroblasts from murine origin [14], which recreated a secondary lung tumor of a metastatic breast malignancy. After 10 days in culture (600 µm in diameter), these 3D structures were exposed to a low or a high concentration of inhalable MNPs and to AMF for 1 h, which resulted in a dose-dependent increase in cell death. The reached temperature was not reported and the contribution of heat for the observed outcome was not excluded, but the authors suggest it to derive from mechanical deterioration. Very recently, Mamica et al. proposed a new in vitro model consisting of a tumor-on-a-chip to study the effects of MHT on treating glioblastoma ( Figure 2) [15]. The tumor was mimicked by using rat glioblastoma cells in 3D configuration in a device integrating microfluidics that guarantees the delivery of nutrients to the tumor. Future developments would include the emulation of the blood vessels in this model. Schematic comparison between magnetic hyperthermia (MHT) in vivo and MHT on a tumor on-a-chip. In vivo, magnetic nanoparticles (MNPs) injected into the tumor are heated up by an alternating magnetic field (AMF). On the tumor-on-a-chip model, MNPs are injected using microfluidics into the central compartment where a 3D tumor structure was formed. The whole chip is then submitted to the AMF. Adapted with permission from Reference [15]. Copyright 2019 Instituto Israelita de Ensino e Pesquisa Albert Einstein.
Ethical constraints impose a limit on the bearable tumor weight/volume for the experimental animal, but most of the in vivo MHT studies report initial tumor volumes ranging from 50 mm 3 [16] to 1.4 cm 3 [17] depending on the experimental animal model (Table A3).  Ethical constraints impose a limit on the bearable tumor weight/volume for the experimental animal, but most of the in vivo MHT studies report initial tumor volumes ranging from 50 mm 3 [16] to 1.4 cm 3 [17] depending on the experimental animal model (Table A3).

Animal Models for In Vivo Studies
The animal models mostly used for MHT studies are rodents, in particular, mice (Table A3). Yet, the efficiency of MHT has also been tested in bigger animals, such as rabbits [17], which can accommodate larger tumors. In many cases, the tumors are generated from xenographs of human cell lines, which is a procedure that requires immunocompromised animals. However, mouse cells can also generate the tumors, and, in this situation, immunocompetent mice may be used to gain insights into the role of anti-tumor immunity in the MHT treatment [18,19]. In fact, there is evidence of intense immuno-stimulation, in particular CD8 + T-cells recruitment and activation, following mild MHT treatment, which may contribute to increased efficiency of the treatment [19].

Parallel Tests in Normal Cells
There are indications of the existence of distinct thermal susceptibility between normal and cancer cells [20,21]. However, this is only true for a certain range of temperatures and it does not exclude the occurrence of side effects mediated by nanomaterials (NMs). Therefore, it is important to test the proposed in vitro MHT strategies in normal cells to understand whether secondary effects should be expected when transitioning to in vivo studies. Only a minority of the studies mentioned in this review actually tested this (see Tables A1 and A2). Fibroblasts, either from mouse origin [22][23][24][25] or from human [26][27][28] origin, are commonly used for this end. For targeted MHT studies, it may be more useful to use a cell line with a lower or even negative expression of the selected target, as this allows to simultaneously prove the selectivity of the treatment [21,25]. This type of study is particularly relevant when the selected target is also widely expressed in normal cells. As an example of the relevance of taking this type of study, Liao et al. functionalized MNPs with D-Galactosamine (D-Gal) to specifically target liver cancer cells, which results in very low levels of cell viability [29]. Even though the selected target, asialoglycoprotein receptor, is known to be widely expressed in normal liver cells as well [30], the authors did not reproduce the test in normal cells and, therefore, the occurrence of cytotoxicity in normal cells cannot be ruled out.

Size, Coating, and Chemical Composition
The size, size distribution, and shape of a particular NM have a clear impact on its magnetic properties [31]. These parameters should ideally be optimized [32] so as to exhibit the highest heating power under a selected AMF frequency [33][34][35], while displaying minimal toxicity if not subjected to an external magnetic field. This optimization also minimizes the need to revise extrinsic parameters, such as the NM concentration or the AMF power.
For MNPs, there is a critical size below which a superparamagnetic behavior is observed. Under an external magnetic field, the atomic magnetic moments of superparamagnetic nanoparticles align along the field direction, which achieves a high magnetic susceptibility [36]. Unlike other types of magnetic materials, once the magnetic field is removed, superparamagnetic nanoparticles behave like a non-magnetic material exhibiting no magnetic memory (i.e., no remanence), which yields stable and very useful colloidal dispersions for biomedical applications [37]. Additionally, the MNPs' size also has clinical and in vivo significance. In fact, MNPs sized above 200 nm are rapidly taken up by the reticulo-endothelial system (RES) and accumulate in the liver and spleen while MNPs sized below 6 nm are filtered by the kidney [38]. Furthermore, there are natural differences between the neovasculature of tumors (defective and leaky) and that from normal tissues with the openings of normal vessels being generally less than 10 nm [39]. These differences pose practical implications since MNPs sized above 10 nm can extravasate and accumulate in the tumor, but not in the normal tissues. Additionally, the slow venous return and lower lymphatic clearance, which are characteristic of tumors, favor MNPs' retention at the tumor site [40]. This extensive leakage and low clearance characteristic of many solid tumors are known as the enhanced permeability and retention (EPR) effect [39]. This seems, however, to fail in the clinical context [41]. At a cellular level, the upper size limit consensually considered for nanoparticle uptake through endocytosis is 500 nm [38,42]. As already mentioned, nanoparticles sized above 200 nm are generally taken up by specialized cells of the RES through phagocytosis, which is a specific type of endocytosis [43].
The Curie temperature (Tc) is an intrinsic characteristic of the magnetic NM that depends on the NM's composition. It is described as the temperature above which the materials become paramagnetic, i.e., the magnetism is lost and the heating stops [47]. This self-regulation of the system's temperature by manipulation of the NM's Tc has been suggested to maintain the temperature of the system within the hyperthermal range [48] in order to decrease the risk of overheating and consequent damage of neighboring (normal) tissues.
The coating is a major contributor for NMs' stability with significant repercussion on their magnetic properties as well as internalization ability and biocompatibility [49], which allows the application of NMs for nanomedicine purposes in general [31]. The coating materials protect the MNPs from oxidation, humidity, and acidity, and create a hydrophilic environment that prevents agglomeration, while allowing for further functionalization ( Figure 3). Additionally, coating can act as a biocompatible shield for the MNPs, which may prevent their opsonization by the RES, consequently, increasing their blood circulation time [50]. Polymers, such as poly(ethyleneglycol) (PEG), poly(lactic-co-glycolic acid) (PLGA), alginate, dextran, and chitosan, are examples of commonly used polymeric stabilizers. Organic non-polymeric stabilizers can also be used for this end, namely oleic acid, stearic acid, and citric acid [51]. More details on the properties of NMs for MHT can be found in recent reviews [31,34,44]. already mentioned, nanoparticles sized above 200 nm are generally taken up by specialized cells of the RES through phagocytosis, which is a specific type of endocytosis [43].
The Curie temperature (Tc) is an intrinsic characteristic of the magnetic NM that depends on the NM's composition. It is described as the temperature above which the materials become paramagnetic, i.e., the magnetism is lost and the heating stops [47]. This self-regulation of the system's temperature by manipulation of the NM's Tc has been suggested to maintain the temperature of the system within the hyperthermal range [48] in order to decrease the risk of overheating and consequent damage of neighboring (normal) tissues.
The coating is a major contributor for NMs' stability with significant repercussion on their magnetic properties as well as internalization ability and biocompatibility [49], which allows the application of NMs for nanomedicine purposes in general [31]. The coating materials protect the MNPs from oxidation, humidity, and acidity, and create a hydrophilic environment that prevents agglomeration, while allowing for further functionalization ( Figure 3). Additionally, coating can act as a biocompatible shield for the MNPs, which may prevent their opsonization by the RES, consequently, increasing their blood circulation time [50]. Polymers, such as poly(ethyleneglycol) (PEG), poly(lactic-co-glycolic acid) (PLGA), alginate, dextran, and chitosan, are examples of commonly used polymeric stabilizers. Organic non-polymeric stabilizers can also be used for this end, namely oleic acid, stearic acid, and citric acid [51]. More details on the properties of NMs for MHT can be found in recent reviews [31,34,44].

Targeting
The overexpression of some surface-receptors in cancer cells compared to normal cells allows for actively targeting cancer cells with MNPs functionalized with a targeting molecule such as an antibody, a peptide sequence, or a ligand protein [52]. This is in contrast with the passive targeting achieved with the EPR effect. The active targeting may result in efficient internalization of targeted MNPs by receptor-mediated endocytosis [53], and it has been described as one of the main factors affecting the binding of MNPs to cells in vitro [54]. The protein adsorption layer (protein corona), formed by the proteins in the medium where nanoparticles interact with cells (e.g., blood proteins in vivo, serum proteins in vitro), is known to contribute to lower adhesion, and, consequently, result in lower uptake. Yet, if a targeting moiety is present on the nanoparticles, the adhesion of the nanoparticles to the cells will fully depend on the target recognition [55]. In this sense, the preferential accumulation (and potential internalization) of targeted MNPs in cancer cells provides a controlled strategy to kill cancer cells with minimal predicted effects in normal cells. Even though recent studies support the superiority of targeted versus non-targeted MHT, both in vitro and in vivo [56,57], others defend that the amount of MNPs delivered by active targeting is insufficient to generate adequate heating at the tumor site [51]. Improvements to targeted MHT efficiency can be achieved by combining it with other therapies [26,[57][58][59][60]. The authors preferred to explore MHT in a mono-therapeutic context while combining targeted with non-targeted MNPs for increased efficiency [21,61]. By adjusting the amount of non-targeted nanoparticles as a magnetic boost for the targeted nanoparticles, it is possible to push up the temperature to the level necessary to induce consistent cancer cell death rates. This strategy was highly effective in vitro, without the need to resort to highly cytotoxic, non-selective, chemotherapeutic agents.
Examples of regularly targeted molecules are the folate receptor and the human epidermal growth factor receptor 2 (HER2) receptor. While the folate receptor was found to be overexpressed in a broad variety of cancers and minimally expressed in normal tissues [62], the HER2 receptor is specifically expressed in some aggressive types of breast cancer [63].

Concentration, Time of Incubation with the Cells, and Nanomaterial Excess Removal
The amount of NM and the time that it is incubated with cells prior to the AMF application usually positively correlate with the uptake of magnetic material, often expressed as pg Fe/cell, which, in turn, contributes to the reached temperature and consequent toxic effects. Comparatively to normal cells, cancer cells seem to display an improved ability to uptake NMs [21,[64][65][66][67]. Even though this behavior cannot be generalized to other cell types rather than the studied ones, it appears as a natural mechanism favoring NMs' accumulation in cancer cells.
It seems hard to establish the Fe loading necessary to achieve efficient MHT, but some studies tried to address this issue. A cancer cell suspension containing a concentration of 2-4 g Fe/L has been reported to generate temperatures between 40 and 45 • C, which causes a 50% drop in cell viability, while 5 g Fe/L resulted in 65 • C and total cell death [68]. However, using high concentration of NM for the MHT tests increases the chances of observing secondary effects, particularly in the in vivo context. A study has suggested that a minimum of 250,000 cells, each loaded with 200 pg of Fe, is required to reach 41.3 • C [13]. Considering the final volume of 0.5 mL, this represents a concentration of 0.1 g Fe/L, which is far below the ones from the previous study. In line with this result, the authors' work on in vitro MHT using a combination of targeted and non-targeted MNPs reported temperatures of 43 to 47 • C and cell viability levels below 20% for cancer cells with a total iron load of 0.31 [61] and 0.65 g Fe/L [21]. When normal cells were treated under similar conditions, a total Fe loading of 0.2 g Fe/L was achieved and the temperature was kept around 41 • C with minimal impact on cell viability [21]. The single cell iron loading for those studies may be found in Table A2. Others have reported elevated MHT efficiency using very low concentration of the luteinizing hormone-releasing hormone (LHRH)-targeted MNPs [69]. The cells containing ca. 15 pg Fe/cell were heated up to 44 • C, which resulted in 5% cell viability. This is a low Fe content when compared with the study by Liao et al. that reported similar viability results but with ca. 365 pg Fe/cell [29]. This discrepancy may result from distinct thermal susceptibility of the cell lines or the different number of cells used in the experiments (5× higher in Reference [69]), which results in overall increased magnetic loading of the sample. Additionally, in Liao et al., the alginate-coated MNPs were incubated only for 30 min with the cells while, in the research from Taratula et al., the poly(maleic anhydride-alt-1-octadecene)-poly(ethylenimine)-poly(ethileneglycol)-coated MNPs were incubated with the cells for 12 h, which suggests an increased uptake capability for alginate-coated MNPs.
Many of the in vitro studies on MHT for cancer treatment do not refer to the concept of intracellular hyperthermia introduced by Gordon et al. [2], in the sense that AMF is applied right after adding the magnetic materials to the cells or without previously removing the extracellular or unbound MNPs. In those studies, the outcome, though sometimes impressive [70] and still dependent on the reached temperature and (time of) exposure to AMF, may be due to the heating of the surrounding environment and, therefore, not necessarily dependent on the direct interaction of MNPs and cells [70][71][72][73][74][75]. On the other hand, Blanco-Andujar and co-workers have recently reported magnetically-induced cell death due to intracellular heating, while excluding the heating effects of the local environment [11].

Administration Route
For in vivo (and clinical) MHT studies, the route of administration of the NMs, along with their coating, contributes to differential distribution and accumulation of the NMs, and a balance needs to be found between the effective tumor treatment and the appearance of side effects due to off-target accumulation in vital organs such as the liver, spleen, and lungs. Many different administration routes have been explored in in vivo MHT studies (see Table A3), but there seems to be a preference for intra-tumoral (IT) and intravenous (IV) administration.
The IT administration of MNPs seems to be generally considered a more invasive procedure than the IV counterpart and it is known to result in uneven distribution of the MNPs in the tumor, which can lead to a significant temperature difference across the tumor [76]. This can be minimized by multiple site IT injection [77] or by magnetic targeting of MNPs after IV injection, i.e., using an external magnet to concentrate the MNPs in the area of interest [76,78]. Magnetic targeting, combined with antibody targeting and the EPR effect, have been reported to allow the tumor to be specifically heated [60]. Among the advantages of IT administration of MNPs are the less complex formulation usually required and that, in general, it results more effectively than the IV route [78]. Additionally, it opens the possibility to treat tumors where MNPs do not accumulate to sufficient amounts after IV injection.
In the case of IV administration of MNPs, a general dilution effect of the blood and the tumor volume should be considered as well as the limited injection volume. Additionally, the majority of the MNPs tends to accumulate in organs such as the spleen or the liver [79], and only a small fraction of the injected material will accumulate in the tumor, which may limit MHT efficiency [80]. Xie et al. confirmed that one single IV injection dose of MNPs does not result in accumulation of enough magnetic material to produce tumor heating [81]. Therefore, six repeated IV injections were performed, which was followed by 15 AMF cycles of 30-min each. This was still not enough to produce a reassuring result. Other authors have previously reported on the challenges of effective MNPs' delivery to the entire tumor with active targeting [53,82]. The heat efficiency of the NM plays a crucial role in this scenario, which means that elevated in vivo MHT efficiency is still possible after IV administration [83]. A major advantage of IV administration of targeted MNPs for MHT is the possibility of equally treating tumor metastasis. Furthermore, while the accumulation of MNPs in the brain is very limited [79], the IV route for the injection of MNPs for treating brain cancer is discouraged. The natural proneness of MNPs to accumulate in the liver after IV injection [82,83] makes this organ an appealing target for MHT cancer therapy. It is interesting to notice, however, that, in most of the studies, the IT administration route is the one selected by the authors [45,77,[84][85][86][87][88].

AMF Power
The AMF parameters to be used for the MHT treatment are usually established in advance before the first in vitro and in vivo studies. For that purpose, the heating performance of the MNPs is routinely assessed by subjecting them to different combinations of field amplitude and frequency for different periods of time [75,89]. The heating power of MNPs is commonly described by the specific absorption rate (SAR, expressed in W·g −1 ), which is also known as a specific loss power. This is the quantification of power dissipation of magnetic NMs under a specific AMF: where C is the specific heat capacity of the sample, m Fe is the iron mass per unit volume, and ∆T/∆t is the initial slope of the temperature (T) vs. time (t) curve [89]. Care must be taken, however, when comparing SAR values because they are affected by a number of parameters such as size, size distribution, chemical composition, and concentration of the NM, but also the amplitude (H) and frequency (f ) of the applied AMF [90]. The product of these two variables, the Hf factor, correlates with the heating power and provides a quantitative index of the heating potential for a given AMF protocol. To avoid damaging the healthy tissues due to electromagnetic radiation (e.g., eddy currents), a threshold Hf of 5 × 10 9 A·m −1 ·s −1 has been established [47,76]. Yet, this threshold is overcome in at least half of the MHT studies addressed in this paper (Table 1). Another factor to be considered is the use of different MHT setups, as the coil dimensions and the range of field amplitude and frequencies may vary, which, therefore, results in distinct SAR values [32]. Additionally, the MHT application protocol may vary between groups. Makridis et al. have addressed the impact of the AMF field conditions on MHT efficiency by testing two different NMs (Cobalt-Fe and Manganese-Fe) under two distinct MHT protocols (single pulse versus multiple pulse) to induce osteosarcoma cell death [91]. The Mn-Fe showed better MHT efficiency than the Co-Fe under the same AMF, which clearly exemplifies the impact of higher SAR values in the MHT efficiency. The heating rate and the heating location were considered the dominant factors to explain the enhanced efficiency of the multiple pulse protocol over the single pulse protocol.

Reached Temperature and Time of Exposure
The reached temperature is a crucial element to be disclosed in an MHT study due to the clear impact it may have in the observed outcome and cell death pathway. Still, nearly 30% of the studies herein addressed do not communicate such information. The temperature reached in an MHT study depends on the characteristics of the NMs and the magnetic field. In turn, the hyperthermia-derived effect will depend not only on the reached temperature but also on the type of cells and the time of exposure to heat (Figure 4) [92]. In in vitro studies, the sample's temperature is usually measured using an optical fiber submerged in the cell culture medium. Infrared thermal imaging cameras are also used both in vivo and in vitro [80,83,93,94], which displaying a color-code image that correlates with the temperature reached in a specific area. Less conventional, and still under development, techniques include, for example, the use of fluorescence anisotropy-based thermoprobes, which are bioconjugates of dyes and proteins with increased thermo-sensitivity comparatively to the individual molecules [95].
At a tissue level, a rise in temperature to around 42 °C results in an increased tumor blood supply, which may be beneficial for the simultaneous delivery of chemotherapeutic agents. This enhances their anti-tumor effect [96]. Also, increased blood flow leads to increased oxygenation, which is a factor known to enhance radio-sensitivity [97]. On the other hand, an augment in temperature to above 42 °C decreases the tumor's blood flow, while increasing blood flow in the normal tissues, which results in a lower dissipation rate and, therefore, a faster temperature rise in tumors, compared to normal tissues [98]. At a cellular level, biophysical and metabolic differences render cancer cells more susceptible to elevated temperatures than normal cells [20]. This fact implies the existence of a temperature range for which it is possible to kill cancer cells with minimal effects in normal cells, which is a major concern when using hyperthermia for cancer treatment.
Studies have demonstrated that the time of exposure to the AMF has an influence on the cell survival rate and clonogenic activity [9,10,75]. However, time only plays a role in case a damaging temperature is reached. In fact, a 1 °C drop within the temperature range of 42.5 to 47 °C can be compensated by duplicating the time of exposure, but, for temperatures below 42.5 °C, the time of exposure needs to be extended significantly more [9].
Court and co-workers found a positive correlation between cytotoxic effects, the reached temperature and the time of exposure, with 45 °C for 30 min inducing a drastic decrease in cell viability (below 10%) [10]. Under milder conditions, between 41 and 43 °C, the cytotoxic effects were intensified by combining the MHT with HSP70 inhibition, either by the HSPA6 gene knock-down, or by inhibiting HSP70 function using 2-phenylethyenesulfonamide. Slightly higher temperatures might be necessary in an in vivo scenario, as complete tumor regression mostly occurs when magnetic thermoablation is performed at temperatures higher than 45 °C [45,83,99]. This sometimes requires a higher NM dose and may result in adverse effects, such as bleeding and infection [99]. Still, some studies reaching temperatures above 50 °C did not report complete tumor regression, which may derive from the short-term exposure to such a high temperature, the bigger size of the tumor, or the different thermal susceptibility of the xenographs [17,84].
The number of MHT cycles also has an impact on the observed outcome, especially in in vivo studies (Table A3), where the number of cycles may vary from 1 to 15. The use of different cells, NM, and MHT protocols, makes it impossible to directly infer the influence of the number of cycles on the MHT efficiency. However, Zhang et al. reported that a lower number of MHT cycles resulted in apoptosis of cancer cells, and, consequently, tumor recurrence. On the other hand, a single cycle producing magnetic thermoablation led to cell necrosis and resulted in complete tumor regression [99]. This suggests that the number of MHT cycles becomes a trivial parameter in the cases where thermoablation is achieved. In in vitro studies, the sample's temperature is usually measured using an optical fiber submerged in the cell culture medium. Infrared thermal imaging cameras are also used both in vivo and in vitro [80,83,93,94], which displaying a color-code image that correlates with the temperature reached in a specific area. Less conventional, and still under development, techniques include, for example, the use of fluorescence anisotropy-based thermoprobes, which are bioconjugates of dyes and proteins with increased thermo-sensitivity comparatively to the individual molecules [95].
At a tissue level, a rise in temperature to around 42 • C results in an increased tumor blood supply, which may be beneficial for the simultaneous delivery of chemotherapeutic agents. This enhances their anti-tumor effect [96]. Also, increased blood flow leads to increased oxygenation, which is a factor known to enhance radio-sensitivity [97]. On the other hand, an augment in temperature to above 42 • C decreases the tumor's blood flow, while increasing blood flow in the normal tissues, which results in a lower dissipation rate and, therefore, a faster temperature rise in tumors, compared to normal tissues [98]. At a cellular level, biophysical and metabolic differences render cancer cells more susceptible to elevated temperatures than normal cells [20]. This fact implies the existence of a temperature range for which it is possible to kill cancer cells with minimal effects in normal cells, which is a major concern when using hyperthermia for cancer treatment.
Studies have demonstrated that the time of exposure to the AMF has an influence on the cell survival rate and clonogenic activity [9,10,75]. However, time only plays a role in case a damaging temperature is reached. In fact, a 1 • C drop within the temperature range of 42.5 to 47 • C can be compensated by duplicating the time of exposure, but, for temperatures below 42.5 • C, the time of exposure needs to be extended significantly more [9].
Court and co-workers found a positive correlation between cytotoxic effects, the reached temperature and the time of exposure, with 45 • C for 30 min inducing a drastic decrease in cell viability (below 10%) [10]. Under milder conditions, between 41 and 43 • C, the cytotoxic effects were intensified by combining the MHT with HSP70 inhibition, either by the HSPA6 gene knock-down, or by inhibiting HSP70 function using 2-phenylethyenesulfonamide. Slightly higher temperatures might be necessary in an in vivo scenario, as complete tumor regression mostly occurs when magnetic thermoablation is performed at temperatures higher than 45 • C [45,83,99]. This sometimes requires a higher NM dose and may result in adverse effects, such as bleeding and infection [99]. Still, some studies reaching temperatures above 50 • C did not report complete tumor regression, which may derive from the short-term exposure to such a high temperature, the bigger size of the tumor, or the different thermal susceptibility of the xenographs [17,84].
The number of MHT cycles also has an impact on the observed outcome, especially in in vivo studies (Table A3), where the number of cycles may vary from 1 to 15. The use of different cells, NM, and MHT protocols, makes it impossible to directly infer the influence of the number of cycles on the MHT efficiency. However, Zhang et al. reported that a lower number of MHT cycles resulted in apoptosis of cancer cells, and, consequently, tumor recurrence. On the other hand, a single cycle producing magnetic thermoablation led to cell necrosis and resulted in complete tumor regression [99].
This suggests that the number of MHT cycles becomes a trivial parameter in the cases where thermoablation is achieved.

Apoptosis or Necrosis?
The ability to produce effective cell death in conditions of mild hyperthermia (41 to 43 • C) is much lower than in thermoablation conditions (at a temperature above 45 • C). Mild hyperthermia can induce apoptosis, but the effects of heat may be reversible due to the induction of HSP expression, which will counteract the heat-induced effects (thermotolerance) [47]. In the case of thermoablation, heat-induced protein denaturation, cytoskeleton, and membrane disruption, and altered deoxyribonucleic acid (DNA) conformation (among other molecular effects) most likely induce a necrotic type of cell death [9]. Table 2 describes the main characteristics of mild hyperthermia versus thermoablation.

Time-Point after MHT
The efficiency of the implemented MHT protocol can be assessed immediately or sometime after the treatment. While the first approach may provide useful information, it does not guarantee the long-term preservation of the observed results. The heat insult may cause a sudden deregulation in cellular metabolism, which can be recovered if the death stimulus is removed and the conditions are favorable [9,11,25]. Sometimes, the actual cytotoxic effect may not be correctly evaluated until hours or even days after the treatment [11,61,100]. This is particularly relevant in studies where apoptotic temperatures are reached, instead of necrotic ones, as some of the surviving cells will be able to multiply and regrow the tumor. As an example, the viability of Jurkat cells exposed to a 1 h-MHT treatment that reached~43 • C increased when the evaluation was performed 72 h, as compared to the 24 h after-treatment [61]. On the other hand, a full kill was observed when the treatment reached~45 • C, which prevented cell recovery and would, expectedly, avoid tumor re-growth. Therefore, even though such treatments are often considered very efficient, there is still a margin for the cancer cells to grow back the tumor when full kill is not achieved.

Most Commonly Used Tests
3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) reduction and trypan blue exclusion are the most frequently used tests to assess cell viability in in vitro MHT studies (Tables A2 and A3). The MTT assay measures the ability of NAD(P)H-dependent oxidoreductase enzymes in viable cells to metabolize the tetrazolium compound to a purple formazan [101]. This provides information of the momentary metabolizing activity, which can be transitorily impaired following the heat insult but can recover in case the conditions become favorable [9], and particularly if the cells exhibit a shorter doubling time [102]. Additionally, care must be taken in the use and interpretation of the MTT assay results since NMs can recognizably interfere with optical detection methods [103]. On the other hand, the trypan blue exclusion assay is based on the ability of viable cells to exclude the dye due to intact cell membranes [104]. Cells that are stained by trypan blue have, therefore, a damaged membrane, which is an irreversible type of cytotoxicity.
In case of in vivo studies, MHT efficiency is often described in terms of suppression of tumor growth or, more desirably, complete tumor elimination, by measuring the size or the weight of the remaining tumor after treatment (Table A3, Figure 5A,B). Taking advantage of the magnetic properties of MNPs, some authors use them as contrast agents, not only for diagnostic purposes but also to evaluate treatment efficiency with magnetic resonance imaging (MRI, Figure 5C) [23,56,105]. Alternatively, generating the tumor with luciferase-expressing cancer cells equally allows tracking tumor size using bioluminescence ( Figure 5D) [106]. While suppressing tumor growth is an encouraging result, only complete tumor elimination can prevent tumor re-growth [19,45,83,87,89]. Therefore, in vivo MHT studies, approaching either mono-therapeutic or combinatorial schemes should aim at complete tumor elimination for more permanent results.
In case of in vivo studies, MHT efficiency is often described in terms of suppression of tumor growth or, more desirably, complete tumor elimination, by measuring the size or the weight of the remaining tumor after treatment (Table A3, Figure 5A,B). Taking advantage of the magnetic properties of MNPs, some authors use them as contrast agents, not only for diagnostic purposes but also to evaluate treatment efficiency with magnetic resonance imaging (MRI, Figure 5C) [23,56,105]. Alternatively, generating the tumor with luciferase-expressing cancer cells equally allows tracking tumor size using bioluminescence ( Figure 5D) [106]. While suppressing tumor growth is an encouraging result, only complete tumor elimination can prevent tumor re-growth [19,45,83,87,89]. Therefore, in vivo MHT studies, approaching either mono-therapeutic or combinatorial schemes should aim at complete tumor elimination for more permanent results.

Remarks and Perspectives
The development of new magnetic NMs and the constant need to develop effective cancer therapies have considerably raised the interest of the scientific community in MHT studies during

Remarks and Perspectives
The development of new magnetic NMs and the constant need to develop effective cancer therapies have considerably raised the interest of the scientific community in MHT studies during the past decade. Even though MHT has been approved by the European Medicines Agency for treating glioblastoma multiforme [97], the use of MHT to clinically treat cancer is still in its very infancy. This may be linked to the fact that different research groups used significantly different MHT protocols, which made it challenging or even impossible to draw comparisons between MHT experiments. Therefore, to help in the future implementation of MHT studies, we compiled the most relevant parameters to consider for the outcome of an MHT assay. These primarily concern the choice of (1) the biological component, (2) the NM, (3) the MHT setup and conditions, and (4) the assessment of the MHT outcome.
The susceptibility of cancer cells to heat is variable and, therefore, a certain MHT protocol may be quite effective in killing certain cells and not others [8]. Similarly, the efficiency of the procedure may differ between in vitro and in vivo settings [10]. Using cellular models that most closely resemble an in vivo scenario should be, therefore, highly supported. The thermal susceptibility of the cell line, the number of cells used in the study, and the 2D or 3D configuration of the cells are details that will impact the results.
The size, the shape, and the coating of the NM are crucial characteristics directly influencing the outcome of an MHT experiment [32,107]. In an in vivo setting, these specifications, together with the presence of a targeting moiety, also affect the NMs distribution, which provides an additional indirect contribution to the MHT outcome [31,44]. Even though NMs that self-regulate the hyperthermia temperature have been suggested as a way to minimize damaging healthy tissues, these were barely tested in vitro [48]. Therefore, more studies are necessary to support their potentiality. In general, biocompatible NMs with increased SAR values are preferred because they are more efficient heat mediators. However, the dependence of SAR on the Hf product brings about the need to control the AMF parameters to stay below the biologically-safe threshold. A fine-tuning is, therefore, necessary to balance the AMF power with the NMs concentration to obtain efficiency while avoiding undesired toxicity.
The outcome of the MHT experiment strongly depends on how and when it is evaluated [61,100]. Time is needed for some types of toxicity to manifest (e.g., apoptosis), but also to investigate whether recurrence (or tumor re-growth) will happen in standard conditions. In this sense, a standardization of the time-point at which the outcome of the MHT experiment would be checked would be an important consideration. This, alone, would immediately indicate whether a certain protocol yields more promising results than another. Importantly, most of the in vivo MHT studies focus on the effectiveness of the treatment by tracking the outcome months after MHT. However, little is mentioned about the fate of the NMs in the body, which is an issue that should be more considered in future in vivo studies.
The difficulty to find effective MHT protocols that could be applied in clinical conditions resulted in the suggestion to use MHT as an adjuvant to other therapies [10,93,94,108]. Most studies show that this results in a more efficient treatment, yet does not exclude the occurrence of side effects that usually accompany the individual therapy.
To summarize, in our opinion, the starting point for the implementation of an MHT (pre)clinical experiment should be the maximum tolerable AMF power threshold. Next, the selection of a biocompatible NM with fine-tuned characteristics, including SAR, would be essential. The protocol should aim at localized heating, using targeting antibodies or magnetic gradients that attract the NMs to a specific location, or focus the AMF in a certain area, which keeps healthy tissues unheated. The NMs developed for this end seem to be, in general, well tolerated but the dependence of the heat production on the NM concentration/payload and the AMF parameters still needs to be established and clarified. For preclinical tests, the biological component should resemble the clinical situation as much as possible. Advantage should be taken on the natural differences between normal and cancer cells, namely their thermotolerance and their magnetic loading ability. Since the future will increasingly rely on the use of human-based, more advanced 3D in vitro models, the assessment of the MHT efficiency should be standardized as the first step to allow the comparison between the outcomes of different MHT protocols, which possibly leads to more competence in the transition from the preclinical to clinical application of MHT.

Conflicts of Interest:
The authors declare having no conflicts of interest.