Characterization of Ferromagnetic Composite Implants for Tumor Bed Hyperthermia

Abstract

Hyperthermia therapy (HT) is becoming a well-recognized method for the treatment of cancer when combined with radiation or chemotherapy. There are many ways to heat a tumor and the optimum approach depends on the treatment site. This study investigates a composite ferromagnetic surgical implant inserted in a tumor bed for the delivery of local HT. Heating of the implant is achieved by inductively coupling energy from an external magnetic field of sub-megahertz frequency. Implants are formed by mechanically filling a resected tumor bed with self-polymerizing plastic mass mixed with small ferromagnetic thermoseeds. Model implants were manufactured and then heated in a 35 cm diameter induction coil of our own design. Experimental results showed that implants were easily heated to temperatures that allow either traditional HT (39–45°C) or thermal ablation therapy (>50°C) in an external magnetic field with a frequency of 90 kHz and amplitude not exceeding 4 kA/m. These results agreed well with a numerical solution of combined electromagnetic and heat transfer equations solved using the finite element method.

I. Introduction

Over the past decades, hyperthermia therapy (HT) has become a recognized adjuvant cancer therapy in multimodality treatment of malignant tumors [1]–[3]. Besides the direct thermal effect on tumor cells, hyperthermia has a sensitizing effect for radiotherapy and chemotherapy [4]–[8] as well as acting as an immunomodulator [9]–[10].

Several heating techniques are used to treat the tumor depending on its size and location in the body. Local hyperthermia induced by microwave radiation is often used to treat tumors close to the skin surface due to limited penetration of microwave radiation, which falls off exponentially into tissue. There are numerous reports of effective treatment for tumors located superficially to a depth of 3–4 cm using external microwave waveguide applicators, but significant heating of overlying tissues is inevitable. Somewhat deeper lying tissue can be heated with a phased-array microwave applicator [11]–[13]. To heat tumors located deep in the body, radio-frequency systems are typically used with a phased-array of antennas concentric to the body, though the longer wavelength leads to a large thermal focus [14], [15]. Regardless, significant heating of normal tissues outside the tumor target is inevitable with all these external heating approaches. As an alternative to electromagnetic radiation for heating deep-seated tumors, ultrasound sources with shorter wavelength and lower attenuation in soft tissue can be used, which make it possible to obtain a significantly smaller thermal focus at depth in soft tissue [16]–[18]. There are also attempts to use combination electromagnetic and ultrasonic radiation [19]. However, approaches using ultrasound are problematic for body regions that contain highly heterogeneous tissues and cavities due to strong refraction and reflection of ultrasound at the boundaries of air and bone inhomogeneities.

Localized heating of tumors can be achieved using internal heat sources placed inside or adjacent to the tumor [20] using ferromagnetic interstitial implants such as needles, rods or spheres. These implants are heated in a sub-megahertz magnetic field [21]–[25] since the ferromagnetic material can produce much higher local heating than the heating induced in tissue directly by the magnetic field [26]. When using small diameter needles or catheters containing ferromagnetic material, a very high-temperature gradient arises at the needle-tissue interface. As a result, tissues directly adjacent to the implanted needle are overheated, while tissues more than 3–4 mm distance from the needle surface do not heat well [27].

In recent years, magnetic nanoparticles have been investigated as heat-generating implant material. To achieve localized heating, the particles are generally concentrated in the tumor region either by injecting nanoparticles into the bloodstream for accumulation in tumor or by directly injecting the magnetic fluid into the tumor [28]–[31]. Since it is challenging to deliver and distribute a sufficiently high concentration of magnetic nanoparticles throughout a tumor, powerful alternating magnetic fields are required to heat a tumor volume sufficiently. Although the performance of nanoparticles for hyperthermia applications is strongly dependent on their magnetic properties in combination with the applied magnetic fields, heating a medium-sized tumor to 42°C can require an alternating magnetic field with at least 100 kHz frequency and at least 10 kA/m strength at a nanoparticle concentration in tumor of about 10 mg/ml [31]. Moreover, it is known that patients are uncomfortable due to direct tissue heating when using alternating magnetic fields with a frequency of more than 100 kHz with a strength of more than 5–10 kA/m [31], [32]. Recently, new approaches for hyperthermia tumor treatment use hybrid scaffolds or stents with incorporated magnetic nanoparticles [33], [34].

A more aggressive form of thermal therapy may be applied by increasing tissue temperature from mild hyperthermia (< 45°C) to thermal ablation (> 50°C) levels to induce immediate cell death [35]–[38]. Although this method is used more often to treat focal disease in liver [39], [40] and prostate [41], there are other successful reports in other organs [42]–[46]. For example, using ablation to treat the mucosal defect after endoscopic resection has been shown to reduce the number of relapses by four times from 21% to 5.2% [46].

We recently reported a multimodality treatment for laryngeal cancer that combines surgery, high dose rate radiotherapy (HDR) and HT, using a custom-formed implant placed intraoperatively into the bed of a resected tumor [47], [48]. The essence of the method is (1) to manufacture an in situ implant that fits the tumor bed; and (2) to use the implant as a solid base to hold carefully positioned catheters for subsequent insertion of an HDR source. The tumor bed-shaped implant allows fixed catheter positioning and accurate calculation and minimization of radiation dose in surrounding at-risk tissues. Moreover, we described the insertion of ferromagnetic particles within the implant to obtain local contact HT of the tumor bed, while applying a biodegradable film containing antitumor or other drugs to the implant surface [49], [50]. In the current study, we assess the feasibility of heating such tumor bed-shaped implants for carrying out either mild HT or more aggressive thermoablation. We aim to heat the tumor bed tissue located close to a resection cavity wall using alternating magnetic field strengths that do not induce significant direct heating of surrounding healthy tissues.

II. Materials and Methods

A. Experimental Setup

The induction heating device was developed at our institution, specially made for preclinical tests in vitro and in vivo. Heating experiments were carried out in a 5-turn solenoid coil measuring 35 cm in diameter and 24 cm in length, similar to that described in [51]. A magnetic field was generated around the region of the implant by feeding the induction coil with a 90-kHz alternating current with a maximum magnetic field strength near the coil center of about 4 kA/m. The magnetic field was measured via the electromotive force induced in a custom-built 1-turn test coil with 1 cm diameter. We used the same magnetic field strength for each experiment in order to directly compare relative output of different implants.

The test chamber used to measure power generation was built as a dual-wall cylinder with an inner diameter of 7 cm. Water was circulated at a controlled temperature (37°C) between the inner and outer walls of the test cell throughout the experiment’s duration. A schematic representation of the measuring cell is shown in Fig. 1a. A spherical model implant was supported in the center of the cell via a vertical glass rod. Four 1-mm diameter glass tubes were placed around the sphere to allow placement of temperature sensors. These tubes could move in the radial direction along slots cut in the cell lid. The guide tubes were fixed in the slots with plastic screws at 0 mm, 2 mm, 4 mm and 6 mm (± 0.5 mm) distances from the implant surface. Optical fiber sensors of a 4-channel thermometer FOTEMP1–4 (Weidmann Technologies Deutschland GMBH, Germany) were inserted into the tubes in the equatorial plane of the implant as shown in Fig. 1b. The error of temperature measurement using optical sensors is ±0.2°C according to the manufacturer. The temperature distribution around each inductively-heated implant was averaged over four experiments, rotating the implant around the axis by an angle of approximately 90° between trials. Before each trial, the system was restored to its initial state with all four sensors at 37.0 ± 0.2°C. It took about four hours for the cell to return to equilibrium after each heating experiment.

Fig. 1.

Experimental setup: (a) Dual-wall measurement cell showing the location of a spherical implant at the center filled with a gel and surrounded by glass tubes for insertion of temperature sensors, and (b) positioning of four fiber optic temperature sensors in equatorial plane of implant.

For modeling tissues with low blood perfusion, we used a MEDIAGEL ultrasound (US) gel (Geltek, Russia) with a viscosity of 18–23 Pa·s.

The implants consisted of a mixture of Speedex putty self-polymerizing plastic (Coltène/Whaledent AG, Switzerland) that hold together 1-mm AISI 52100 steel spheres (Resurskomplekt LLC, Russia) used as thermoseeds. The putty material is widely used to make casts for prosthetics in dentistry and will be referred to as putty in the remaining document. We manufactured the implants with 2-cm and 3-cm diameter with a steel mass fraction of 30%, 50% and 70%. Note that, in a clinical setting, the implant is formed intraoperatively by mechanically filling the tumor resection cavity with the soft plastic pre-mixed with steel spheres, which then hardens into the tumor bed shaped implant after a few minutes. More details of the surgical procedure can be found in reference [48].

B. Computational Modeling

The alternating magnetic field and temperature distribution in and around the implant were numerically solved using the finite element method software COMSOL Multiphysics 5.1 (COMSOL LLC, Russia). We used the software modules AC/DC and Heat Transfer, which solve low-frequency magnetic field, temperature modeling problems. Due to the expected symmetry of the experimental setup, we used an axisymmetric geometry for modeling (Fig. 2) that mimics the experimental setup in Fig. 1. The computational model of the experimental setup includes a cylindrical region (air) that confines the full integration region surrounding the phantom and coil, with 1 m diameter and 1 m height that surrounds the cylindrical phantom domain with equal height and diameter of 7 cm (blue region). Due to the very high viscosity of ultrasound gel (18–23 Pa·s), we modeled this region as a solid. The induction coil is modeled using five separate turns (C1-C5) with 35 cm diameter, where each turn consists of copper tubing with 1 cm diameter and a wall thickness of 1 mm.

Fig. 2.

Computational model of the implant and phantom shown in Figure 1 as well as the surrounding 5 coils. The coils generate the magnetic field required to activate the steel thermoseeds immersed in the 2–3 cm implant.

The thermostatic chamber effect was modeled in a boundary condition at the cell wall using a constant temperature of T_therm = 37°C to simulate core temperature. The initial temperature in the phantom region was set at 37°C. The surface of the Environment region was set at constant temperature T_env = 25°C. Magnetic Insulation boundary condition was also set at this surface. Table I shows the material properties used in the simulations. The physical properties for steel and air were retrieved from COMSOL Multiphysics material’s library. Electrical conductivities of the US gel and Speedex putty were set to be equal to zero as they are much lower compared to electrical conductivity of steel. The thermal conductivities of US gel and Speedex putty were chosen by fitting to our experimental data. The steel relative permeability at 90 kHz was set equal to 100 as previously in [48]. No hysteresis was added to the calculation scheme, as this steel has low remanence and coercivity. More computational details are available as supplementary material.

TABLE I.

Material properties used in the computational model of the heating implant

	Air	US gel phantom	Steel sphere	Speedex putty
Relative permeability, μr	1	1	100	1
Relative permittivity, εr	1	2	1	3
Electrical conductivity, σ (S/m)	0	0	1·107	0
Density, ρ (kg/m3)	1.2	1100	7800	1300
Specific heat capacity, c (J/kg/K)	1000	2000	450	2000
Thermal conductivity, k (W/m/K)	0.01	0.55	50	0.3

To estimate the heat generated by the implant, first we calculated the heat produced by one thermoseed and then estimated the average heat produced in the entire implant volume. The corresponding numerical model consisted of a single seed inserted in the putty material which was immersed in a 90-kHz magnetic field generated within the 5-turn induction coil. We then integrated the power loss density (PLD, W/m³) in this one seed implant at steady state, resulting in the calculated power generated by one thermoseed P₁. The heat generation of the whole implant was then calculated using PLD_i = NP₁/V_i, where N is the number of seeds in the implant (Table 2), V_i is the implant volume, and i is the index referring to the spherical implant. This heat source was then incorporated in the heat transfer equation and solved numerically in COMSOL. For this calculation, we assume independence of the magnetic field near any thermoseed on its environment as a first approximation and then considered the estimated value as a fitting parameter.

TABLE II.

Composite properties of the heating implant used in the numerical simulations

φm	N	φv	ρ, kg/m3	c, J/kg/K	k, W/m/K
0.3	533a	0.067	3250	1535	0.36
	1800b
0.5	1143a	0.143	4550	1225	0.44
	3857b
0.7	2240a	0.280	5850	915	0.62
	7560b

^a2-cm implant

^b3-cm implant

Each thermoseed in an implant is located in a magnetic field produced by both the induction coil and all other steel seeds. This fact can be taken into account in terms of the effective relative permeability (μ) using, for example, the Maxwell model [52] described by the following equations:

$${\mu }_i={\mu }_p\frac{1+2{\phi }_v{\beta }_{\mu }}{1-{\phi }_v{\beta }_{\mu }},{\beta }_{\mu }=\frac{{\mu }_s-{\mu }_p}{{\mu }_s+2{\mu }_p},$$ (1)

where ϕ_υ is the volume fraction of steel in the composite implant (Table 1), and μ_s and μ_p are the relative permeabilities of steel and putty, respectively. Note that we dropped the index r for simplicity. Using equation (1) and the data from Table 1, we can calculate the implant relative permeability for steel mass fractions of 0.3, 0.5 and 0.7, yielding μ_0.3 = 1.21, μ_0.5 = 1.48 and μ_0.7 = 2.12, respectively. The effective electric conductivity (σ) of the implants can be estimated in a similar way:

$${\sigma }_i={\sigma }_p\frac{1+2{\phi }_v{\beta }_{\sigma }}{1-{\phi }_v{\beta }_{\sigma }},{\beta }_{\sigma }=\frac{{\sigma }_s-{\sigma }_p}{{\sigma }_s+2{\sigma }_p}.$$ (2)

using the electric conductivities of steel (σ_s) and putty (σ_p). For all mass fractions under consideration (0.3, 0.5 and 0.7), equation (2) gives σ_i = σ_p = 0 S/m. The value of skin-layer thickness estimated using effective magnetic permeabilities and electric conductivities of the implants at mass fraction of steel of 0.7 is about 30 m if σ_p = 1·10⁻¹⁰S/m. That is, all thermoseeds inside the implants may be considered as being placed in the same magnetic conditions.

The density and heat capacity of each implant were estimated using the formulas:

$$\begin{gathered} {\rho }_i={\phi }_m{\rho }_s+\left(1-{\phi }_m\right){\rho }_p, \\ {c}_i={\phi }_m{c}_s+\left(1-{\phi }_m\right)c_p, \end{gathered}$$ (3)

where φ_m is mass fraction of steel in the composite implant; ρ_s and ρ_p are the density of steel and putty densities; and c_s and c_p are steel and putty specific heat capacities, respectively. To determine the effective thermal conductivity (k) of the implant, the Maxwell scheme was used again, according to which

$$k_i=k_p\frac{1+2{\phi }_v{\beta }_k}{1-{\phi }_v{\beta }_k},{\beta }_k=\frac{k_s-k_p}{k_s+2k_p},$$ (4)

where k_s and k_p are steel and putty thermal conductivities, respectively. Table 2 presents the thermophysical parameters of composite implants estimated using formulas (3) and (4) as well as the number of seeds in the different implants using the formula N = V_i/V₁, where V₁ is the volume of a sphere with 1-mm diameter.

III. Results

Fig. 3 presents simulation results for the magnetic field generated by the model induction coil with units of militesla (mT). Note that color scales are different for each figure to provide optimal visualization of the field gradients in each region. The magnetic field within the central region of the 35 cm diameter coil is quite uniform, and field inhomogeneity within the 7 cm diameter phantom region does not exceed (4.62–4.48)/4.55 = 0.031, i.e., about 3%. The magnetic field strength near the coil center is 3.6 kA/m, corresponding to a magnetic flux density of 4.6 mT. The field strength of the equivalent experimental setup was 3.6±0.4 kA/m, which correlates well with the numerical simulations.

Fig. 3.

Modelling of magnetic field in the coil and phantom: (a) integration region of the computational model; (b) physics-controlled mesh; (c) magnetic flux density distribution in the coil; (d) magnetic flux density distribution in the phantom region.

Fig. 4 shows the simulation results for the heat generation in composite implants with different mass fraction of steel. Fist, we calculated the heat power released by one 0.5–1.5 mm diameter thermoseed. The heat absorbed by unit volume of the implant is seen to increase with the increase of thermoseed size (Fig. 4a) and mass fraction (Fig. 4b). This result is in good agreement with the results of preliminary published experiments in [47].

Fig. 4.

Modeling of absorbed power per unit volume induced by a 90-kHz magnetic for different: (a) diameter (ϕ) thermoseeds (PLD for a single thermoseed); (b) mass fractions (φ_m) of the composite implants filled with 1-mm thermoseeds.

Table 3 shows fitted PLD values for the implants of two different diameters filled with 1-mm steel thermoseeds.

TABLE III.

Simulated power loss density for different mass fractions and implant diameters

φm	Implant Diameter (cm)	PLD (W/m3)
0.3	2	1.22·105
	3	1.65·105
0.5	2	2.85·105
	3	3.4·105
0.7	2	5.55·105
	3	6.3·105

Fig. 5 shows experimental and simulation results for heating a 2-cm diameter implant having different mass fractions of 1-mm thermoseeds. The phantom used was the ultrasound gel. As seen in Fig. 5b, these power absorption rates would result from an applied field strength of approximately 4 kA/m. The magnetic field strength measured near the center of the induction coil used for our experiments was 3.6 ± 0.4 kA/m.

Fig. 5.

Temperature distribution around a 2-cm heated composite implant immersed in ultrasonic gel for 1-mm thermoseeds at different mass fractions φ_m. The different curves show the temperature profile at a given distance (d) between the temperature sensor and implant surface. Dots are used for experimental data and solid lines for numerical simulations.

Fig. 6 shows experimental and simulation results for a 3-cm diameter composite implant with the same mass fraction of thermoseeds as the 2-cm diameter composite implant. However, as seen in Table 3, the Heat generation rate of a 3-cm diameter implant is 35%, 20% and 15% higher than a 2-cm diameter implant.

Fig. 6.

Temperature distribution around a 3-cm heated composite implant immersed in ultrasonic gel for 1-mm thermoseeds at different mass fractions φ_m. The different curves show the temperature profile at a given distance (d) from the implant surface to the temperature sensor. Dots are used for experimental data and solid lines for numerical simulations.

IV. Discussion

Definitive treatment of oncologic disease is among the most challenging issues in modern medicine. Especially in the case of advanced malignant neoplasms, there is a high risk of relapse leading to death due to the inability to completely remove all malignant tumor cells from the adjacent tissue. Besides, the classic surgical treatment usually leads to loss of function of the affected organ as a result of its radical removal. Sometimes the loss of physiological functions of the body may be associated with the loss of other important body functions. A vivid example of social disability is speech loss in the treatment of laryngeal cancer. For this reason, there is increased interest in developing new organ-preserving methods for oncologic treatment of critical body organs and treatment sites.

In this study, we focused on heating a custom-built composite ferromagnetic surgical implant intended for insertion in a tumor bed. These implants are formed using a self-polymerizing plastic mass mixed with small ferromagnetic thermoseeds. Due to the flexibility of these implants, they can be used to heat tumor bed even around irregular shape organs. This publication demonstrates the heating capability of a spherical composite implant consisting of biocompatible polymer containing a homogenous distribution of steel thermoseeds with 30, 50 or 70% mass fraction with both numerical simulations and experimental measurements.

Fig. 4a shows that the efficiency of induction heating of a composite implant increases with the size of steel thermoseeds. Use of seeds with diameter < 1 mm requires increasing the concentration of the seeds to the point where the lower percentage of polymer produces a fragile structure of the composite implant. On the other hand, using a thermoseed size >1 mm can lead to excessive heating of each seed and higher risk of degradation of polymer in contact with the thermoseeds. Our experiments show that thermoseeds with 1-mm diameter provide reasonable uniformity of heat production within the polymer matrix and sufficiently high power absorption to provide temperature rise of the composite implant due to induction heating at practical magnetic field levels less than 4 kA/m. For this reason, all experiments represented in Fig. 5 and Fig. 6 were performed using 1-mm thermoseeds.

As seen in Figs. 5 and 6, composite implants are able to heat the 2- and 3-cm implant surfaces and adjacent at-risk tissue to either moderate hyperthermia temperatures (e.g 30% mass fraction), or to higher thermally ablative temperatures if desired (with 50% or more steel fraction). It should be noted that such heating was carried out in a 90-kHz magnetic field with strength not exceeding 4 kA/m, which is sufficiently gentle electromagnetic conditions to minimize direct tissue heating outside the implant region. Note that these favorable results were achieved for the ultrasound gel phantom that mimics low-blood perfusion tissues only.

Let us now discuss the reason for difference of PLD for 2 and 3 cm diameter implants at the same mass fractions of steel (see Table 3). The results in Fig. 5b were obtained under the assumption that each thermoseed was placed in the same magnetic field generated by the coil and not affected by the other steel seeds. However, each steel seed in an implant is located in a magnetic field produced by both the induction coil and all other steel seeds. To explain the higher values in Table 3 compared to values estimated from Fig 5 we use the equation (1). Effective magnetic permeability of the implant, and therefore magnetic flux density B_m inside the implant increases with increasing fraction of magnetic filler. Thus, the power loss density, which is proportional to $B_m^2$, increases with increasing steel seeds concentration. To explain the different values of fitted heat power density for 2 and 3 cm diameter implants with equal mass fractions of steel it should be understood that the Maxwell model is valid only for the case of homogeneous mixture of two components. However, homogeneity is impaired near the implant surface. Taking into account that the implant volume is proportional to the cube of its dimension and its surface proportional to the square of its dimension, we conclude that the larger the volume of the implant, the more accurate the Maxwell model is, and the more pronounced is the concentration of magnetic flux in the implant.

One problem that could result from the different heating rates of different diameter implants is that large diameter regions of a complex elongated shape implant may heat surrounding tissue to higher temperatures than around smaller diameter regions. That difference would be accentuated in tissue with high perfusion due to the higher required field strengths. One of the possible ways to reduce the uneven heating of an implant of a complex shape can be the use of a balloon filled with either ferromagnetic liquid [53] or nonmagnetic liquid metal alloy, for example, based on indium and gallium [54]. Future studies will investigate the imbalance of power deposition within irregular shaped implants and whether it can be offset by increasing the implant thermal conductivity to improve thermal conduction equilibration of temperature throughout irregular shaped-implants.

V. Conclusion

This work investigates the heating capability of ferromagnetic implants made from various concentrations of 1-mm diameter steel thermoseeds immersed in a biocompatible polymer, as intended for insertion into a surgical resection cavity following removal of tumor. Experimental results agree with numerical simulations which demonstrate that 2–3 cm diameter spherical polymer implants filled with 30–70% (by mass) steel thermoseeds are easily heated to temperatures that allow either traditional mild hyperthermia or high-temperature thermal ablation of adjacent tissue-equivalent material by immersing the implant in a magnetic field with amplitude not exceeding 4 kA/m at 90 kHz.