Quantitative Photothermal Characterization with Bioprinted 3D Complex Tissue Constructs for Early‐Stage Breast Cancer Therapy Using Gold Nanorods

Abstract

Plasmonic photothermal therapy (PPTT) using gold nanoparticles (AuNPs) has shown great potential for use in selective tumor treatment, because the AuNPs can generate destructive heat preferentially upon irradiation. However, PPTT using AuNPs has not been added to practice, owing to insufficient heating methods and tissue temperature measurement techniques, leading to unreliable and inaccurate treatments. Because the photothermal properties of AuNPs vary with laser power, particle optical density, and tissue depth, the accurate prediction of heat generation is indispensable for clinical treatment. In this report, bioprinted 3D complex tissue constructs comprising processed gel obtained from porcine skin and human decellularized adipose tissue are presented for characterization of the photothermal properties of gold nanorods (AuNRs) having an aspect ratio of 3.7 irradiated by a near‐infrared laser. Moreover, an analytical function is suggested for achieving PPTT that can cause thermal damage selectively on early‐stage human breast cancer by regulating the heat generation of the AuNRs in the tissue.

The accurate prediction of heat generation on tumor tissue is indispensable for clinical plasmonic photothermal therapy (PPTT). An analytical function to predict the temperature variation is suggested by analyzing heat generation using various 3D tissue construct and computational biophysics analysis. Also, selective thermal damage on early‐stage human breast cancer is confirmed using bioprinted 3D complex tissue constructs.

1. Introduction

As a new therapeutic strategy for cancer treatment, plasmonic photothermal therapy (PPTT) using metallic nanostructures is now considered to be a potentially effective means of inducing tumor destruction, owing to the high absorption coefficient of the plasmonic bands and high spatial selectivity, which allows the effective heat to be preferentially direct to the tumor region.^{[ 1} , ² , ³ , ⁴ , ⁵ , ⁶ , ^{7 ]} In PPTT, localized photothermal heating is induced by the surface plasmon resonance (SPR) effect resulting from the interaction of nanoparticles with incident light at the resonant frequency. The effectiveness of PPTT depends on the abilities of the photothermal transducer to absorb specific wavelengths of incident light strongly and convert resonant energy into heat energy efficiently.^{[ 6} , ^{7 ]} Recently, gold nanoparticles (AuNPs) have become preferred as photothermal transducers, owing to their superior physical properties, such as strong light absorption, simple surface chemistry, enhanced biocompatibility, and the ease of tuning the localized SPR peaks in the near‐infrared (NIR) region (0.7–1.1 µm).^{[ 8} , ^{9 ]} Various AuNPs, such as nanoshells, nanorods, nanocages, nanocrosses, and nanostars, have been reportedly utilized using NIR light.^{[ 10} , ¹¹ , ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁶ ]

It has been demonstrated that PPTT using AuNPs can elevate the temperature of the biological tissue to 42 °C, which is sufficient to cause apoptosis and ultimately coagulative tumor necrosis.^{[ 10} , ¹¹ , ¹² , ¹⁴ , ^{15 ]} Owing to the efficacy of PPTT, many researchers have focused on understanding the optical, biological, and thermal interactions between AuNPs and biological tissues at cellular and organismal levels.^{[ 5} , ¹¹ , ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ^{21 ]} These studies provided valuable data, for PPTT research to progress to preclinical stages; however, several important obstacles to clinical application remain. Notably, the reproducibility and effectiveness of PPTT must be guaranteed regardless of the intratumoral microenvironment, which is dependent on the cancer‐cell line for in vitro models, since cancer treatment is commonly performed repeatedly because of the aggressiveness and resistance of cancer.^{[ 22} , ^{23 ]} Unfortunately, the low reproducibility and reliability of PPTT processes and the difficulty of in vivo temperature measurement around tumor sites have not been resolved. Nevertheless, accurate prediction of the tumor‐site temperature changes, based on the fact that the plasmonic heat generation is a function of laser power density and AuNP concentration, helps counteract these weak points. The investigation of PPTT in clinical practice has been attempted mostly from computational approaches that exploited various mathematical models developed to govern optical‐thermal responses of laser‐irradiated tissue.^{[ 19} , ²¹ , ²⁴ , ^{25 ]} However, more realistic methods that can mimic the optical and thermo‐physical properties of the real tissue surrounding tumors and PPTT mechanisms within tissue are also strongly needed, to predict the in vivo heat generation and to understand the heat‐transfer mechanism and tumor tissue response.

Herein, we present a strategy for quantitatively characterizing the PPTT heating conditions and applying thermal controls for successful tumor destruction, which is verified by modeling the PPTT process. Bioprinted 3D complex tissue constructs comprised of processed gel obtained from porcine skin (pSG) and human decellularized adipose tissue (hDAT) with cancer cells were developed to characterize the process and was compared to polymethylsiloxane (PDMS) constructs to model the AuNP heat generation in tissue. Moreover, we suggest a computational model of multi‐layered human breast tissue to characterize the selective thermal cancer damage in human in vivo environments. As the target cancer type, early‐stage breast cancer located relatively near the skin surface^{[ 26 ]} was selected because it coincides well with the concept of PPTT and direct irradiation with NIR light. Light of these wavelengths can penetrate 1‐cm below the skin surface without the use of invasive devices in normal soft human tissues, which are highly scattering and weakly absorbing in the 0.6–1.4 µm wavelength range.^{[ 27 ]} The in vitro 3D breast tissue model (the bioprinted 3D complex tissue construct) was fabricated by casting two different materials, pSG and hDAT, and layering them to mimic the optical and physical properties of individual skin layers. Each pSG and hDAT layer represented the dermis, which is composed of collagen and elastin fibers in an amorphous matrix of mucopolysaccharides, and the hyperdermis, which is a fatty subdermal layer,^{[ 28} , ^{29 ]} respectively. The photothermal properties of citrate‐coated gold nanorods (AuNRs) with an aspect ratio of 3.7 irradiated by an 808 nm laser were also characterized in each layer. Moreover, a comparative analysis between this model and one based on porcine breast skin tissue (pBS), which is known to be the most similar tissue to human skin in its histological and biochemical properties,^{[ 30} , ³¹ , ^{32 ]} was conducted to characterize the AuNR heat generation within the human breast tissue. Assuming an in vivo environment, a computational biophysics analysis of PPTT for superficial breast cancer was performed using radiative transport theory with the delta‐P1 approximation, Pennes’ bio‐heat equation, and the Arrhenius equation. Consequently, an analytical function describing the effect of PPTT on early‐stage human breast cancer was obtained using various mathematical approaches. This function can be employed to determine the AuNR amount and laser power with respect to exposure time at various tissue depths that are required to cause irreversible tumor damage in vivo. Our findings show that the in vivo heat generation of AuNPs can be estimated under given conditions, enabling the design of PPTT treatment regimes to provide fatal, irreversible damage to cancer cells selectively, owing to the localized AuNP heat generation.

2. Results

2.1. Rational Design of 3D Tissue Constructs and Experimental Setup

The 3D tissue constructs were designed to match the cross‐sectional photothermal distribution in the depth direction and to induce a moderate temperature rise in a specific target region clinically termed as hyperthermia. Because an infrared (IR) camera can only provide information about the material surface, the heat generated by AuNRs in tissue typically cannot be measured accurately. Therefore, in this work, the tissue constructs were made in a vertically truncated form to image the cross‐sectional heat distribution in the depth direction. The constructs contained 2‐mm‐diameter hemispheres that were treated as cancerous tissues implanted with AuNRs (Figure  1a). To measure the temperature variation in a laser‐irradiated AuNR solution inside the tissue constructs, a collimated 5‐mm‐diameter laser beam was aligned to be incident perpendicular to the surface of each tissue construct, and the thermal distribution and heating of specific concentrations of AuNRs were measured using the IR camera after 300‐s of laser exposure (Figure S1, Supporting Information). To identify the laser characteristics that help define the power density, temperature variation, and exposure time of use, the laser power, thermal distribution, and sustainability of heat generation were characterized (Figure S2, Supporting Information). As the laser beam intensity in a defined diameter has a Gaussian distribution, to identify the thermal distribution in the area around the AuNRs, the temperature point value is shown separately in each IR thermograph. Each temperature represents the average of temperature point values within the 2‐mm‐hemisphere area (π mm²) on the top surface of the relevant tissue. Consequently, to characterize the correlated synergistic effects of laser power and amount of AuNRs on the heat generation, seven different laser power densities (18.72 (D _P1), 23.40 (D _P2), 28.08 (D _P3), 32.76 (D _P4), 37.44 (D _P5), 46.80 (D _P6), and 56.16 W cm⁻² (D _P7)) and eight different AuNR concentrations (0.46 × 10⁸ (C ₀₀), 0.85 × 10⁸ (C _X), 1.70 × 10⁸ (C ₀), 2.55 × 10⁸ (C ₁), 3.40 × 10⁸ (C ₂), 4.25 × 10⁸ (C ₃), 5.10 × 10⁸ (C ₄), 5.95 × 10⁸ (C ₅), and 8.48 × 10⁸ (C ₆) AuNRs/µL) at various depths were tested. The exothermic temperatures ranged from 3 to 50 °C, depending on the combination of these factors and the type of material.

Figure 1.

Characterization of the photothermal effect of AuNRs irradiated by an 808‐nm laser in the 3D PDMS constructs. a) Illustration of a sample construct aligned perpendicular to the laser for imaging the cross‐sectional heat distribution of AuNRs in the tissue construct. A representative image of a 3D PDMS construct and a thermal map of AuNRs in the tissue construct exposed to the laser for 300 s obtained using an IR optical microscope are shown in the insets. b) IR thermographs and temperature variations showing the AuNR heat distributions in 3D PDMS constructs obtained with laser power densities D _P1, D _P5, and D _P7 and the fixed concentration C ₃. The direction and diameter of the laser are marked on the image. The thermal distribution of AuNRs within a 2‐mm‐diameter hemisphere in the tissue construct was used to calculate the temperature generated by laser irradiation by averaging all points displayed. The thermal maps were derived from the thermographs and the temperature variation is displayed below in each case. c) AuNR thermal distributions and temperature variations obtained for AuNR concentrations C ₁, C ₃, and C ₅ and the fixed laser power density D _P5. d) Characterization of the interrelated effects of the amount of AuNRs and laser power on the AuNR heat generation in the 3D PDMS constructs. The mean temperature variation in the area of 2‐mm‐diameter‐AuNRs is represented as the mean ± SD of five consecutive measurements and is exponentially fitted with a solid line. Three independent experiments were conducted for each individual set of conditions.

2.2. Photothermal Effect of AuNRs in PDMS

To characterize the AuNR heat generation regulated by laser irradiation, the laser must reach the AuNRs without power loss caused by light scattering, reflection from the material surface, or absorption by the material. Therefore, to measure the heat generated by the AuNRs in response to NIR laser irradiation, 3D constructs were first fabricated using PDMS, which has low light absorption^{[ 33 ]} and scattering coefficients^{[ 34 ]} at NIR wavelengths. With D _P1 and C ₃ at a depth of less than 1 cm, there were no significant differences (SE ± 0.72) between the temperatures at depths of 1, 3, 7, and 10 mm (data not shown), and when the AuNRs were 10‐mm from the surface rather than 1‐mm, the temperature was reduced only by 0.78 ± 0.04%. To characterize the interrelated effects of the amount of AuNRs and laser power on the AuNR heat generation, the concentrations C ₀₀–C ₅ and laser powers D _P1–D _P7 were tested. AuNRs having an aspect ratio of 3.7, which have the highest absorption at NIR wavelengths,^{[ 35 ]} were used, and they were coated with citrate to red‐shift the absorption spectrum closer to 808 nm than that of the commonly used cetyltrimethylammonium bromide‐coated AuNRs (Figure S3a, Supporting Information). Moreover, each AuNR concentration was confirmed by measuring the optical density of the AuNRs to ensure accuracy (Figure S3b, Supporting Information).

In addition to determining the type of laser and AuNR, the photothermal effects of the AuNRs irradiated with the 808‐nm laser were characterized using 3D PDMS constructs (Figure 1). The temperatures around the irradiated AuNRs and their thermal distributions across the skin construct were measured using IR thermographs (Figure 1a). The thermograph of the area outlined by the dotted line in the photograph show the thermal distribution around the AuNRs within the hemisphere surface, and the ambient temperature generated by laser radiation of the AuNRs was calculated by averaging all the points displayed. Furthermore, the 3D thermal distribution around the hemispheres shows the thermal diffusion of the AuNRs into the surrounding tissue. With the fixed AuNR concentration C ₃, the temperature was first measured while varying the laser power density (Figure 1b). As expected, the temperature immediately after laser irradiation increased rapidly with increasing laser power and stabilized within 100 s. The temperature was strongly elevated in the hemisphere and increased by more than 46 °C with the laser power density D _P7, whereas there was no temperature increase on the surface, implying the PDMS construct surface does not absorb the laser. Next, the temperature variation was measured while varying the AuNR concentration and holding the laser power density at D _P5 (Figure 1c). The temperature increased, in response to the laser irradiation, in proportion to the AuNR concentration; the increase was by more than 42.0 °C when the concentration was C ₅.

Consequently, the AuNR heat generation was characterized by the laser intensity and amount of AuNRs, as shown in Figure 1d. The absolute heat generated by the AuNRs without heat absorption or loss by the material was measured in the 3D PDMS constructs. Our results suggest that the AuNR heat generation upon laser irradiation can be controlled by adjusting the AuNR concentration and laser power density. Additionally, the laser intensity and amount of AuNRs needed to increase the temperature by a certain degree can be determined, and the temperature increase when a given laser power and amount of AuNRs are used can be predicted.

2.3. Photothermal Effect of AuNRs in Construct Materials

To observe the photothermal effects of laser‐irradiated AuNRs that pass through the dermal layer and determine the heat generated on the dried skin surface in response to laser absorption, pSG was developed, and its composition and mechanical properties were analyzed (Figure S4, Supporting Information), confirming that the tissue construct comprising pSG had biophysical properties as a workable natural compound similar to those of gelatin, a form of partially denatured collagen. As collagen is the main component of the dermis, and gelatin gel is commonly used as a dermis substitute,^{[ 36} , ^{37 ]} pSG is a suitable model for the dermal layers. By analyzing the cross‐sectional heat distribution of the AuNRs using IR thermographs, the temperature variations and thermal absorption in the pSG construct containing AuNRs were then measured using D _P1, D _P5, and D _P7 with C ₃ (Figure  2a). The results show that, as when a PDMS construct was used, the temperature rapidly increased within 20 s, stabilized after 100 s, and was strongly elevated in the hemisphere in proportion to the power density by up to 17.7 °C with D _P7. However, unlike when a PDMS construct was used, in the thermal distribution of the pSG construct cross‐section, measured for various laser powers (Figure 2a), heat generation was observed from not only the AuNR upon laser irradiation but also from the surface because of laser absorption. Moreover, the heat generated by the AuNRs in the hemisphere at a depth of 7 mm in the pSG construct and the resulting heat generated on the pSG construct surface increased in proportion to the laser power density. However, the heat generated on the pSG construct surface by laser absorption and by irradiation of the AuNR varied independently as functions of the laser power and their distributions overlapped at a depth of 4 mm in the pSG construct under the given conditions.

Figure 2.

Characterization of the heat generation of AuNRs irradiated by an 808‐nm laser in the pSG a–d) and hDAT e–h) tissue constructs. a) Photograph of a pSG tissue construct with the laser direction marked and IR microscopic thermographs showing the AuNR heat generation in the pSG constructs with power densities of D _P1, D _P5, and D _P7 and a fixed AuNR concentration of C ₃. The thermal distribution over the 2‐mm‐diameter area of AuNRs outlined with a dotted line was used to calculate the temperature variation of AuNRs irradiated by the laser. The white scale bars indicate 7 mm. A representative thermal map of the cross‐section of the pSG construct exposed to the laser for 300 s is inserted. b) Temperature variations showing the AuNR heat distributions in the pSG constructs with AuNR concentrations C ₁, C ₃, and C ₅ and the fixed laser power density D _P5. c) Characterization of the interrelated effects of the amount of AuNRs and the laser power on the AuNR heat generation in the pSG constructs. d) The temperature variations in the pSG constructs with tissue depths of 2, 4, and 7 mm was represented as mean ± SD of five consecutive measurements. e) Temperature variations showing the AuNR heat generation and thermal distributions caused by the 808‐nm laser in hDAT constructs with laser power densities D _P1, D _P3, D _P5, D _P6, and D _P7 and the concentration fixed at C ₁. f) AuNR heat distributions in the hDAT constructs with AuNR concentrations C _X, C ₀, C ₁, C ₂, and C ₃ and the laser power density fixed at D _P7. g) Characterization of the interrelated effects of the amount of AuNRs and laser power on the AuNR heat generation in the hDAT constructs. Each mean temperature variations in 2‐mm‐diameter area of AuNRs are represented as a mean ± SD of five consecutive measurements, and values are linearly fitted to the solid line. h) Temperature variations in the hDAT constructs with tissue depths of 3, 5, and 7 mm are represented as mean ± SD of five consecutive measurements. Three independent experiments were conducted, and one experiment is shown as a representative: C ₁/D _P5. The maximum temperature variation is indicated in each thermograph alongside the tissue depth.

After characterizing the AuNR heat generation with various laser powers, the temperature was measured while varying the concentration from C ₁ to C ₅ with D _P5 (Figure 2b). The results show that the temperature was increased by 8–16 °C with increasing concentration, whereas the surface temperature increased by 10 °C regardless of AuNR concentration (Figure 2b). Consequently, the AuNR heat generation was analyzed with respect to two regulatory factors: the laser power and the amount of AuNRs in the pSG constructs (Figure 2c). The temperature around the AuNRs in the pSG construct was reduced by an average of 62 ± 1.2% SE compared to that in the PDMS constructs. These findings suggest that laser power reduction caused by surface absorption, light scattering, and reflection as it penetrates into the pSG construct decreases the AuNR heat generation. Furthermore, the heat generated on the tissue surface due to laser absorption can affect the internal heat generation of the AuNRs in the tissue. To investigate this, the temperature was measured while varying the depth, distance from the tissue surface to the AuNR clusters inside the tissue (Figure 2d). The AuNR heat generation in the pSG constructs significantly decreased with increasing depth, whereas the temperature in the PDMS construct increased by 33 °C regardless of depth. Remarkably, with the AuNRs in the pSG constructs at depths greater than 4 mm, there was less heat generation than in the PDMS constructs. However, with the AuNRs only 2‐mm from the surface, the temperature increased by 42 °C, which is more than in the PDMS construct under the same conditions. This result shows that the heat generated by AuNRs in the pSG construct decreased when less laser power reached the AuNRs, owing to the laser energy loss caused by heat generation on the surface. However, when the AuNRs were sufficiently close to the surface, the surface heat generated by laser absorption caused the AuNRs to emit more heat in the pSG construct. In summary, our findings suggest that first, as the laser beam was absorbed by the tissue surface, heat was generated on the tissue surface, and less laser power reached the AuNRs in the tissue to generate secondary heat. Moreover, the two types of heat generation were independent when the nanoparticles were deep within the skin, but the heat generation on the tissue surface significantly influenced the heat generation of the AuNPs within the tissue when they were close to the tissue surface. Therefore, depth is another important factor in determining the degree of heat generation of the AuNRs inside the tissue.

To investigate PPTT for early‐stage breast cancer, which is spread throughout the fibrous and adipose tissue that comprise the main mass of the breast,^{[ 38 ]} a 3D hDAT tissue construct of was made to characterize the photothermal properties of AuNRs in human adipose tissue (hAT). The hAT obtained from a patient was first decellularized and then solubilized to generate a reproducible biomaterial (Figure S6a, Supporting Information), and this was followed by molding the 3D tissue form (Figure S8, Supporting Information). To quantitatively assess the hAT decellularization process, extracellular matrix (ECM) conservation was confirmed by hematoxylin and eosin staining as well as Masson's trichrome staining, and the ECM and DNA contents of the hDAT were measured and compared to those of native adipose tissue (Figure S7, Supporting Information). The lack of visible nuclei in the staining results and the increased prevalence of ECM components compared with native adipose tissue indicate that the hAT could be effectively decellularized into hDAT. Next, the temperature variations and heat distribution within the hDAT construct were calculated using IR thermographs according to the AuNR concentration, laser power density, and depth. The hemisphere size and depth of the hDAT construct were the same as those of the other tissue constructs. Consequently, photothermal characterization of AuNRs embedded in the hDAT construct was conducted first with C ₁ and laser power densities D _P1–D _P7 (Figure 2e), then with D _P7 and AuNR concentrations C_X –C ₃ (Figure 2f). The results show that the AuNR heat increased considerably around the hemisphere in proportion to the AuNR concentration and laser intensity. The temperature with the concentration fixed at C ₁ increased up to 4.8 °C with D _P7, and with the laser power density fixed at D _P7, it increased up to 7.5 °C with C ₃. Moreover, the heat generated by the hDAT construct surface caused by laser absorption increased up to 4.2 ± 0.1 °C with D _P7 and varied with the laser intensity.

In summary, the photothermal effect of AuNRs in the hDAT construct was characterized, and the AuNR heat generation was found to be linearly proportional to the incident laser intensity and AuNR concentration (Figure 2g). Furthermore, to investigate the AuNR heat generation changes with the depth of the AuNRs beneath the skin, in addition to the laser absorption changes with depth, the temperature was measured as the depth of the AuNR cluster from the hDAT construct surface was varied (Figure 2h). As with the results discussed above, these results indicate that the temperature increased more with increasing proximity of the AuNRs to the surface of the hDAT construct, reaching 5.5 ± 0.2 °C with C ₁ and D _P5 at a depth of 3 mm and 3.3 ± 0.2 °C at a depth of 7 mm. Alongside the heat generated by the laser‐irradiated AuNRs, the surface heat of the hDAT construct also increased up to 5.23 ± 0.5 °C with a depth of 3 mm and 1.95 ± 0.2 °C with a depth of 7 mm. Consequently, compared with the pSG constructs, the AuNR heat generation was reduced, indicating greater laser absorption by the high water content of the hDAT construct. However, the heat generated on the tissue surface caused by laser absorption, as with the pSG construct, also significantly affected the AuNR heat generation within the hDAT construct when the AuNRs were closer to the hDAT construct surface.

2.4. Photothermal Effects of AuNRs in pBS

pBS tissue, which is known to be similar to human skin because of its histological and biochemical properties,^{[ 30} , ^{31 ]} was used to characterize the AuNR heat generation in human‐like complex tissue to validate the breast‐tissue model based on comparison with in vivo breast skin. Figure  3 shows the temperature variation and cross‐sectional heat distribution of AuNRs within the pBS calculated using IR thermographs under various AuNR concentrations and laser power densities. The specimen thickness, length, and hemisphere size were the same as for the other tissue constructs. Consequently, photothermal characterization of AuNRs in the pBS was performed using laser powers D _P1–D _P7 and the fixed concentration C ₃ (Figure 3a), followed by concentrations C ₁–C ₅ and the fixed laser power density D _P5 (Figure 3b). The results show that the thermal response of the AuNRs was similar to those of the AuNRs in the other tissue constructs at first, but the degree of the response differed. For C ₃, the temperature increased from 5.2 ± 0.1 °C to 14.4 ± 0.2 °C as the laser power density increased from D _P1 to D _P7, whereas the temperature with D _P5 increased from 8.2 ± 0.1 °C to 11.4 ± 0.2 °C as the concentration increased from C ₁ to C ₅. Moreover, the heat generation of the pBS surface varied with the laser power density, but not with the concentration.

Figure 3.

Characterization of AuNR heat generation in pBS tissue. a) Porcine breast skin tissue was cut to a depth of 5 mm and a width of 14 mm, and 2 µL of the AuNR solution with concentration C ₃ was injected into the hemisphere 7‐mm away from the tissue surface. IR thermographs showing the AuNR thermal distribution in the porcine tissue with the defined concentration and laser power densities D _P1, D _P3, and D _P5. The white scale bars indicate 7 mm. b) IR thermographs and temperature variations showing the AuNR heat distribution in the pBS with AuNR concentrations C ₁, C ₃, and C ₅ and the laser power density D _P7. c) Computational analysis of cross‐sectional temperature variations of AuNRs 7‐mm beneath the surface of porcine breast skin tissue performed for comparison to the results in Figure 3a. d) Characterization of the interrelated effect of the amount of AuNRs and laser power on the AuNR heat generation in the pBS. The mean temperature variation in the area of 2‐mm‐diameter‐AuNRs is represented as mean ± SD of five consecutive measurements and values are linearly fitted to the solid line. Three independent experiments were conducted for each individual set of conditions. e) Precise control of the emitted temperature generated by laser irradiation in the pBS.

Next, to evaluate the reliability of the results, a multi‐layered porcine breast‐skin model was constructed to perform computational biophysics analysis (Figure 3c and Figure S10: Supporting Information). The simulation was conducted under the same conditions for direct comparison with Figure 3a. The graphical trends and ratio variations in the simulation data agreed well with the experimental data. Quantitatively, the simulation data were 15.1 ± 1.6% lower than the experimental data based on the average values, and 6.2 ± 1.1% lower based on the maximum peak values. These differences resulted from the fact that the simulation data exhibited narrower bell‐shaped curves, resulting in average temperatures lower than those obtained experimentally. We consider that the geometric dimensions of each tissue component (Figure S10, Supporting Information) and the optical and thermo‐physical properties used in the computations (Table S3, Supporting Information) might not correspond exactly to those of the amorphous natural porcine tissue structure. Small differences in the individual properties would have affected the thermal transfer and light transport behaviors in the porcine tissue. Nevertheless, the results demonstrate that the computational biophysics analysis for PPTT is reliable for predicting AuNR heat generation according to the applied conditions.

Finally, the photothermal properties of AuNRs embedded in pBS, as a human‐like model, were characterized while varying the incident laser power density and AuNR concentration (Figure 3d). The AuNR heat generation was linearly proportional to these two regulatory factors. Interestingly, the temperature increase of the AuNRs in pBS was between those of the pSG construct and the hDAT construct. This result is sufficiently predictable because porcine skin tissue consists of complex components, including collagen and adipose. Consequently, this result is compatible with the previous results obtained using the pSG and hDAT constructs. Furthermore, Figure 3e shows that the temperature increase can be finely controlled for thermal cancer damage by adjusting the laser power density and the AuNR concentration. In summary, our findings suggest that AuNR heat generation in vivo is predictable and can be precisely regulated to achieve the appropriate temperature.

2.5. Computational Biophysics Analysis for Superficial Breast Cancer

We performed computational biophysics analysis targeting superficial breast cancer to characterize PPTT quantitatively using AuNRs in an in vivo environment and to elucidate the optical and thermal interactions among the laser, AuNRs, and biological tissues (Sections S9–S13, Figure S9, Tables S1 and S2: Supporting Information). We focused on the temperature and thermal damage distributions in the tumor and normal tissues during PPTT by using an NIR laser with an 808‐nm wavelength, a 5‐mm‐diameter beam, and a 300‐s exposure time and different AuNR concentrations and laser power densities (Figure  4a,b). The calculated results for the different AuNR concentrations (C ₁–C ₅) with the laser power density as D _P1 and laser power densities (D _P1–D _P5) with the AuNR concentration at C ₃ are presented in Figure 4c,d, respectively, which depict the temperature profiles along the laser‐beam axis. On the laser‐irradiated skin surfaces, no significant temperature changes appeared between concentrations (maximum 1.5 °C), while the temperature increases with increasing laser power density (maximum 3.8 °C). Meanwhile, the tumor temperature variation (ΔT) increases linearly with increasing AuNR concentration and laser power density (Figure 4e).

Figure 4.

Computational biophysics analysis of PPTT for superficial human breast cancer. a) Schematic of PPTT for superficial human breast cancer and b) computational model of superficial human breast cancer used in this work. Temperature profiles along the laser‐beam axis at a laser exposure time of 300 s for various c) AuNR concentrations and d) laser power densities. e) Temperature increase curve. f) Temperature distributions in the superficial human breast‐cancer model for the conditions yielding the highest temperature increase in each set of variations (C ₅/D _P1 (left) and C ₃/D _P5 (right)). Transient temperature profiles obtained with a laser exposure time of 300 s for various g) AuNR concentrations and h) laser power densities. Transient thermal damage fraction profiles obtained with a laser exposure time of 600 s for various i) AuNR concentrations and j) laser power densities. Thermal damage distributions at 300 and 600 s with conditions representative of k) the concentration variations, C ₄/D _P1 and C ₅/D _P1, and l) the laser power density variations, C ₃/D _P3 and C ₃/D _P4.

Figure 4f shows the temperature distributions in the breast‐cancer model for the conditions providing the highest temperature increases (C ₅/D _P1 (left) and C ₃/D _P5 (right)). As expected, the highest temperature occurs in the tumor region, and the heat is diffused into the normal tissue radially. On the skin surface, the temperature increased to 38.5 °C with C ₅/D _P1 and 40.6 °C with C ₃/D _P5, whereas the maximum temperature within the tumor (within the black boundary line) was 52.1 °C with C ₅/D _P1 and 59.2 °C with C ₃/D _P5. The transient temperature profiles averaged for the whole tumor region, which was located 7‐mm below the skin surface, are shown in Figure 4g,h for all AuNR concentrations and laser power densities, respectively. All transient temperature curves display similar trends despite the temperature differences; the temperature increased sharply until about 25 s, then it reached a steady state at about 100 s.

In this computational biophysical analysis, the thermal damage fraction (F _d) is a function of time used to describe the cell damage accumulated over time (Section S12, Supporting Information). Therefore, we calculated and observed the transient thermal damage fractions averaged over the whole tumor for all AuNR concentrations and laser power densities, respectively, for 600 s (Figure 4i,k). At 300 s, only two conditions (C ₃/D _P4 and C ₃/D _P5) provided thermal damage fractions greater than 0.6, which indicates sufficient denaturation and injury induced by the temperature elevation throughout the tumor region (Figure 4b). For these two conditions, the thermal damage fraction was very low at early laser exposure times (<50 s), when the tumor temperature nearly attained its maximum (about 51.7 and 53.8 °C for C ₃/D _P4 and C ₃/D _P5, respectively); however, it increased steeply thereafter. At 600 s, three additional PPTT conditions (C ₅/D _P1, C ₃/D _P2, and C ₃/D _P3) provided irreversible damage. These findings indicate that significant cellular damage caused by thermal effects was not immediately generated by a high temperature (50–60 °C). Thus, apart from temperature, the laser exposure time is also important for achieving irreversible tumor damage during PPTT. Hence, a lesser temperature increase (45–50 °C) can also be expected to cause irreversible damage with a long exposure time.

Figure 4j,l shows the thermal damage distributions for two representative conditions after exposure for 300 and 600 s, respectively. Three results for each different trend are shown. For C ₄/D _P1, significant thermal damage is not observable at 300 s. At 600 s, the averaged F _d is close to 0.6; however, only the local tumor damage is evident. For C ₅/D _P1 and C ₃/D _P3, although local tumor damage was apparent at 300 s, almost complete thermal damage was evident at 600 s. Interestingly, despite sufficient temperature elevation on the periphery of the normal tissue (outer part of the 0.5 mm margin from the black boundary line), collateral damage was not observable in either case. This phenomenon could be explained by the fact that healthy cells are less sensitive than cancer cells to thermal damage and a lack of long‐term exposure to temperatures greater than their physiological levels.^{[ 19} , ^{39 ]} For C ₃/D _P4, complete thermal damage of the tumor was evident at 300 s; however, the thermal damage area extended into the normal tissue at 600 s.

2.6. Characterization of AuNR Heat Generation for PPTT with Bioprinted 3D Complex Tissue Construct

To propose PPTT clinical treatment guidelines for early‐stage breast cancer treatment, the heat generation of AuNRs having an aspect ratio of 3.7 irradiated by an 808‐nm laser was first characterized as a function of the AuNR concentration and laser power density. The temperature variations were measured using PDMS constructs, which did not undergo laser‐light absorption or scattering, even with six different laser intensities and AuNR concentrations (Table S4, Supporting Information). Polynomial interpolation^{[ 40 ]} was then used as a numerical approach to approximate a complex surface representing temperature variations (Figure  5a). The order of the polynomial equation that best fit the experimental results was determined to be three, including two input variables, the AuNR concentration (C, #/µL), and the laser power density (D _P, W cm⁻²). Consequently, we obtained the following polynomial equation:

$$\begin{array}{ccc}& & f_{\mathrm{AuNR}}\left(C,D_{\mathrm{P}}\right)=7.598\times {10}^{-5}{C}^3-1.925\times {10}^{-11}{C}^2{D}_{\mathrm{P}}\hfill \\ & & -4.617\times {10}^{-3}{C}^2-1.721\times {10}^{-18}C{D}_{\mathrm{P}}^2+3.564\times {10}^{-9}C{D}_{\mathrm{P}}\hfill \\ & & +3.334\times {10}^{-2}C+1.744\times {10}^{-26}{D}_{\mathrm{P}}^3-3.52\times {10}^{-18}{D}_{\mathrm{P}}^2\hfill \\ & & -4.483\times {10}^{-9}{D}_{\mathrm{P}}+0.567\hfill \end{array}$$ (1)

Figure 5.

Mathematical approach for characterizing AuNR heat generation as a function of AuNR concentration, laser power density, and tissue depth. a) Numerical approach for predicting the ΔT of the AuNRs emitted by an 808‐nm laser according to the AuNR concentration and laser power density. The surface shows the polynomial interpolation of the ΔT variations, and the blue circles represent the experimental results shown in Table S4 (Supporting Information). b) Temperature reduction rate according to depth for the tissue constructs made of various materials compared with that of PDMS, for which had no change in AuNR heat generation with depth and was used as a reference with the temperature reduction rate defined as one.

The temperature variation was calculated as a function of both regulatory factors by using Equation (1). The 3D surface shown in Figure 5a represents the temperature variation with respect to the concentration and laser power density where 30 data points were used to evaluate the polynomial interpolation, and the blue circles correspond to the experimental results in Table S4 (Supporting Information). The mean of the errors between the polynomial interpolation results obtained using Equation (1) and the experiment results for the temperature variations in Table S4 (Supporting Information) was 0.69, and the standard deviation was 0.74. Consequently, this equation can be used to estimate the values of the two variables necessary to achieve the required AuNR heat generation or to estimate the AuNR heat generation, given a combination of the two variables.

In this study, AuNR heat generation was also shown to depend on the thermo‐physical and optical properties of each tissue component, which affect the thermal transfer and light transport behavior, respectively. Therefore, the temperature variations varied with depth within the tissue. For numerical application, the temperature variations were characterized as functions of AuNR depth according to tissue constructs made of various materials (Figure 5b). The temperature in the PDMS construct was set to one, and the relative ratios were calculated for the other materials. As mentioned previously, because the AuNRs were situated deep within the tissue, the heat generation decreased dramatically. In addition to considering the photothermal characteristics of each material, computational biophysics analysis was conducted to calculate the temperature reduction rate (α), which is represented by a red dotted line according to the depth of AuNRs (x) in human breast tissue at which physiological temperature (37 °C), metabolic, and blood perfusion heat sources were considered (Equation S7, Tables S1 and S2: Supporting Information). As a result, the following natural exponential function was obtained:

$$\alpha =2.11e^{-0.18x}$$ (2)

The temperature reduction rate represents the degree to which the light energy reaching the AuNRs in the tissue is reduced by the optical and thermophysical properties of the medium (e.g., the tissue component and water content). It determines the light attenuation and thermal diffusion in the tissue, as well as the heat transferred from surface absorption to the internal heating of AuNRs, resulting in increased heat generation. In summary, AuNR heat generation in vivo can be defined by multiplying the temperature variation as determined by the combination of two regulatory factors by the temperature reduction rate according to the depth of the AuNRs within the tissue. As a result, we obtained the following thermal prediction equation:

$$f\left(\mathrm{\Delta }T\right)=\alpha ·f_{\mathrm{AuNR}}\left(C,D_{\mathrm{P}}\right)$$ (3)

Our findings suggest that it is possible to predict the temperature variation at any depth beneath the skin surface under given conditions, including laser power density and AuNR concentration. Thus, PPTT can be performed such that selective thermal cancer damage can be caused by regulating the heat generation of AuNRs injected into a breast‐cancer mass located near the skin surface.

Finally, bioprinted 3D complex tissue constructs were fabricated to verify our thermal prediction and cellular thermal damage. The commonly used human epithelial breast‐cancer cell line, MCF‐7, was used as the cancer model.^{[ 41 ]} The mixed layer of the hDAT and MCF‐7 cells containing a 2‐mm‐diameter hemisphere concave at its center (1‐mm thick) and the hDAT layer (3–7 mm) were sequentially printed using a customized 3D bioprinter and a pSG layer (0.2–1 mm), which was added and finally inverted. Each layer represents the breast‐cancer mass, dermis, and hypodermis, in order. The AuNRs were loaded on the hemisphere concave in the hDAT and MCF‐7 as mixed layers for irradiation. The laser was irradiated using a 5‐mm‐diameter beam from the bottom perpendicular to the tissue construct. Hence, we can easily modify the depth by changing the height of the hDAT layer (Figure  6a and Figure S6b: Supporting Information). Because the depth of the AuNRs is a major factor for PPTT, we precisely adjusted the depth with bioprinting of each layer while fabricating the AuNR loading site centrally concave on the top layer. After laser irradiation, the temperature variation was measured from the top surface of the bioprinted 3D complex tissue construct using an IR thermograph (Figure 6c). Correlated with our thermal prediction, the temperature variation decreased sharply with increasing depth, and the temperature variation in the bioprinted 3D complex tissue construct was between those of the pSG and hDAT constructs. At a depth of 3 mm, the temperature increased by 9.9 ± 0.6 °C, but only a 5.4 ± 0.3 °C increase was observed at a depth of 7 mm under C ₃/D _P5, and the temperature variation could be adjusted by controlling the concentrations of AuNRs or laser power density (Figure 6d).

Figure 6.

Thermal prediction and cellular thermal damage in a bioprinted 3D complex tissue construct. a) Schematic perspective and top views of bioprinted 3D complex tissue constructs. b) Schematic of relationship between temperature variation and thermal damage of cancer cells. The dotted line shows the boundary of 100% thermal damage. c,d) IR thermographs (C ₃/D _P5) and temperature variations for the C ₃/D _P5 and C ₆/D _P7 conditions with 3‐, 5‐, and 7‐mm tissue depth represented as mean ± SD of five consecutive measurements. Three independent experiments were conducted for each individual set of conditions. e) Analysis of live/dead MCF‐7 cells by microscopy. The 3D complex tissue construct with 7‐mm depth was stained with calcein AM (live, green) and EthD‐1 (dead, red). White dotted line indicates the AuNR‐loaded hemisphere edge. f) Cellular thermal damage with D _P7 laser power density for 300‐s irradiation as a function of the distance from the AuNR hemisphere edge depicted in Figure 6a (top view).

To select the optimal laser power density according to the concentration of AuNRs and the depth of the breast‐cancer mass for clinical application, the ratio and range of thermally damaged cancer cells based on temperature variation should be first investigated. The laser power that causes cell damage precisely targeted to the cancer site provides the optimal condition for PPTT. An excessive intensity will damage the nearby normal tissue, and insufficient irradiation will lead to cancer‐treatment failure. To identify the optimal temperature variation for PPTT, the thermal damage of MCF‐7 cells in the bioprinted 3D complex tissue construct was calculated by counting live and dead cells according to their distances from a hemispheric edge. The dead cells observed before laser irradiation were excluded from the calculation. At the optimal temperature variation, only cancer cells nearest the hemisphere edge will be dead. Hence, thermal damage will be shown as 100% at 0‐mm distance and will decrease sharply with distance. In the case of low temperature variation, thermal damage will fail to reach 100% even at a 0‐mm distance, whereas the thermal damage of cancer cells will be maintained at 100% away from the edge at a high temperature variation (Figure 6b). As expected, the excessive thermogenic conditions damaged cells outside the targeted hemisphere edge where the AuNRs were inserted. The higher the temperature rise, the farther the damaged area from the edge (Figure 6e). During laser irradiation, there was a collapse of tissue near the AuNRs because of cellular and matrix damage. The MCF‐7 cells within an area of 3.50‐mm from the edge were fully damaged under an 8.10 × 10⁸ AuNRs/µL concentration with the D _P7 laser power density, which caused a temperature rise of 7.7 °C. A temperature variation of 5.1 °C with 4.05 × 10⁸ AuNRs/µL and the D _P7 laser power density resulted in complete MCF‐7 cell damage in the area 1.75‐mm distant from the edge. However, an increase in temperature of 4.5 °C caused only 87.8% thermal damage near the edge, meaning that the optimal temperature variation for PPTT was between 4.5 and 5.1 °C (Figure 6f) which agrees with previous reports.^{[ 11} , ¹² , ^{14 ]} These results show that the temperature variation of the AuNRs is always higher than that at the surface, and a temperature lower than 45 °C does not induce skin burns, although the skin is exposed to that temperature for a long period.^{[ 8} , ^{42 ]} Consequently, the optimal laser power for PPTT, which induced a 4.5–5.1 °C increase in cancer tissue, did not cause any damage to the skin.

In summary, we fabricated multi‐layered bioprinted 3D complex tissue constructs and demonstrated that an optimal laser power suitable for specific cancer treatment can be calculated considering AuNR concentrations and the depth of the cancer mass. However, the bioprinted 3D complex tissue construct showed a smaller temperature increase than the human tissue simulation in Figure 5b, suggesting that more factors must be considered for fabricating a better skin construct. For example, the stratum corneum having low moisture content may facilitate a rise in temperature, and the blood circulation system may protect the skin construct by heat redistribution.^{[ 43} , ^{44 ]} We surmise that the temperature difference between the simulation and bioprinted 3D complex tissue construct is caused by the high water content of hDAT layer and heat loss into the surroundings. As a further study, development of a better 3D skin tissue construct that includes the stratum corneum and a blood circulation system is needed to more accurately investigate temperature increases caused by PPTT in human skin tissue.

3. Discussion

Compared with current PPTT strategies for selective cancer treatment, our method provides a thermal prediction equation for estimating the temperature variation in biological tissue. The characterization of PPTT heating conditions and effective thermal control of an ensemble of AuNRs is shown to allow successful tumor destruction in vivo by evaluating its performance using computational biophysics and 3D tissue constructs made of skin components. An equation was developed to represent the effects laser power density, particle concentration, and temperature reduction rate by depth, and this is expected to help overcome critical issues that create hesitation with regard to the clinical application of PPTT.

Nonetheless, the plasmonic photothermal properties of AuNPs clearly depend on their sizes and shapes, ^{[ 45} , ⁴⁶ , ⁴⁷ , ^{48 ]} which affect the localized SPR peak (λ _peak) and the absorption cross‐section (σ _a). Therefore, these factors should be additionally considered to provide more comprehensive PPTT evaluations in various environments. Theoretically, the heat generation capability of plasmonic NPs is directly related to σ _a, and the sample heat generation (Q _sample) is generally quantified using σ _a (i.e., Q _sample = σ _a·N·I·V, where N is the number density of AuNPs; I is the laser intensity; and V is the sample volume).^{[ 46 ]} It means that more heat is generated as the absolute value of σ _a increases. Based on this, σ _a may be critical factor to controlling the temperature variation according to the other variable changes, as shown in Figure 5a. It will lead to the height change in the y‐axis direction depending on the nature of the plasmonic NPs. Hence, the temperature variation value in the biological tissue must be characterized according to σ _a of various plasmonic NPs that is determined by structural variables such as the volume, shape, and material constituents, and environmental variables such as the photodegradation and shape deformation of the NPs induced by high‐intensity laser radiations.

Moreover, the optical spectrum and λ _peak are highly sensitive to the aspect ratio (AR) for AuNRs, thickness for nanoshell, and diameter for nanosphere (i.e., the large shift in λ _peak when varying the AR, thickness, and diameter).^{[ 47} , ^{48 ]} The λ _peak shift requires a change in the laser wavelength used for irradiation (λ _laser), which eventually leads to variation in the light attenuation mechanisms in the biological tissue. For instance, λ _peak of AuNR shifts from 640 to 850 nm when the AR is increased from 1.1 to 4.4, such that λ _laser at λ _peak for a selected AuNR becomes a factor that changes the laser penetration rate into the tissue, as represented by the temperature reduction rate in Figure 5b; this leads to a slope change in the x‐axis direction in this figure. Therefore, the temperature reduction rate must be characterized according to λ _peak, as selected by the chosen type of AuNP. Future research should enable various AuNPs to be used in PPTT depending on the nature of the cancer, and by considering σ _a and λ _laser at λ _peak, the thermal prediction equation can be obtained for clinical application.

Wilhelm et al. surveyed the NPs delivery efficiencies by conducting a multivariate analysis employing the literatures related to the NPs delivery to the tumor which were published in 2005–2015, and they revealed that only 0.7% (median) of the injected dose of NPs reached the tumor site through intravenous injection. ^{[ 49 ]} Hence, AuNP delivering system through blood vessels is not effective for AuNPs transport. Therefore, we suggest the fabrication of the microball containing a defined amount of AuNPs encapsulated by a biomaterial that is thermo‐degradable at 40–45 °C. A microball having a size that is compatible with injection using a syringe (<1 mm) can prevent rapid diffusion in the tissue prior to treatment and induce accurate heat generation by the injected AuNPs. Because the heat is generated by the AuNPs, injecting AuNPs within the tumor is more efficient than injecting them into the surrounding tissue, and this helps to minimize damage to normal cells. Additionally, thermally damaged tumor‐derived tumor antigens may activate immune cells, leading to a further human immune response for complete tumor eradication.

In our study, we used a collimated 5‐mm‐diameter laser beam with an area of AuNRs of >2 mm in diameter to ensure uniform laser irradiation of the target area. However, the homogeneity of laser power density is important for reaching the targeted temperature within a precise cancer tissue volume to ensure uniform damage of an area of cancer cells. Therefore, we plan to add a beam homogenizer to our system in a future study.

4. Conclusions

AuNP heat generation in human tissue is sensitively dependent on the AuNP amount, the induced laser power, and the tissue constituents, making it necessary to clearly understand the interrelated relationships among these variables for successful PPTT development. In this study, the process was first described by the amount of heat that could be generated as a function of the amount of AuNPs, dependent on changes in laser power. We discovered the amount of heat loss that occurred with depth depended on the specific composition of the skin tissue. Then, we estimated the amount of heat that was ultimately generated by specific amounts of AuNPs in the tissue. To further investigate the heat‐generating properties of AuNPs in human‐like complex tissues and to validate the human breast‐tissue model, 3D‐printed multi‐layered tissue constructs were developed using porcine breast tissue. Moreover, for the comprehensive application of PPTT, a mathematical approach to modeling the heat generation of the AuNPs was presented alongside a computational prediction model for targeting superficial human breast cancer to quantitatively elucidate the optical and thermal interactions among the laser, AuNRs, and biological tissue variables for selective thermal cancer damage. Consequently, our results imply that the heat generation along the depth of the tumor containing AuNRs can be precisely governed by controlling of the amount of AuNRs and laser power. This was demonstrated by confirming the selective thermal damage of cancer cells in the tissue. PPTT combined with direct irradiation of NIR light can be applied to superficial breast cancer as well as other specific cancer tumors typically located within the penetration depth of NIR light (e.g., skin and thyroid cancers). Furthermore, this advancement will make a significant contribution to the clinical application of PPTT and will have important implications for both specialists in this field and their patients. However, to predict the heat generation of AuNPs in the tissue more accurately, it is necessary to develop better 3D complex tissue constructs that can simulate the optical, mechanical, and structural properties of human skin tissue, including blood vessels and multiple dermal layers. For precise printing of such multilayer structures with keratinocytes, connective tissues and an extracellular matrix composed of collagen fibrils, microfibrils, and elastic fibers, it is necessary to improve the resolution performance of the 3D printer nozzle. Furthermore, the development of microballs containing a defined number of AuNPs for direct injection into the cancer tissue is needed. Such efforts will practically contribute to the clinical application of PPTT for cancer treatments and the advancement of tissue engineering and biofabrication.

5. Experimental Section

Computational Biophysics Analysis for Superficial Breast Cancer

The computational biophysics analysis of PPTT for superficial breast cancer consisted of the following three steps. First, the light distribution of the laser beam employed for superficial illumination of the breast tissue was determined using radiative transport theory with the delta‐P1 approximation (Section S10, Supporting Information). Second, the temperature distributions within the multi‐layered breast tissue were predicted based on Pennes’ bio‐heat equation (Section S11, Supporting Information). Third, the thermal damage to the tumor and surrounding normal tissues was calculated based on the temperature elevation history using the Arrhenius equation (Section S12, Supporting Information). Following these steps, the PPTT procedures for superficial breast cancer was computationally modelled by using the continuum approach and explicitly applied a multi‐layered 3D symmetric FE model of the breast tissue (Section S9, Supporting Information). All of the PPTT computations for each AuNR concentration and laser power density were conducted using the commercial FE software COMSOL Multiphysics (Version 5.1, COMSOL Inc., Palo Alto, CA), and the details are given in Section S13 in the Supporting Information.

Preparation of pSG Construct

The detailed process used to produce the pSG is shown in Figure S4 (Supporting Information). Briefly, porcine skin chilled for three days after death was obtained from a butcher shop in Jeongup, Republic of Korea. The skin was washed to remove dirt, hair, and dead skin cells from the surface. It was then chopped into sections with dimensions of about 10 × 10 cm. The chopped skin was boiled in water for 30 min at 100 °C. Then, the hot skin was ground up (Hi‐Mixer HM‐350 Hyun Dae Co. at 1000 rpm for 30 min) before cooling down to 50 °C until all of the solid material was completely gone. Finally, it was filtered (stainless steel filter 180 mesh/82 micron welded sus 304, Duda Co., USA) to remove the hair roots and surface matter. The viscous skin tissue gel was poured into a mold and subsequently cured at room temperature. The sealed pSG construct was exposed to radiation (cobalt 60 gamma‐ray, MDS Nordion, Canada, performed at Korea Atomic Energy Institute) at 10 kGy h⁻ for sterilization.

Preparation of hDAT Construct

Human adipose tissue (hAT) samples were provided by a patient undergoing incision of the lower abdomen with informed consent and approval from the Catholic University of Korea Seoul Saint Mary's Hospital. The samples were frozen at −80 °C and processed according to the physical and chemical steps previously described.^{[ 50 ]} The decellularization and hDAT construct fabrication process is shown in detail in Figure S6 (Supporting Information).

Preparation of pBS

The pBS parts refrigerated for one day after death were brought from a butcher shop (Geumcheon Livestock Wholesale Center) in Daejeon. The skin was first frozen at −20 °C for 30 min for cutting into the desired size, as shown in Figure 3a. After cutting 5‐mm‐high, 14‐mm‐wide sections, a 2‐mm‐diameter, 1‐mm‐deep indentation was made on the cross‐sectional surface 7 mm away from the skin edge by stamping with a rounded‐end 2‐mm‐diameter iron rod, and the AuNR solution with a defined concentration was injected into the indentation.

Fabrication of the 3D Tissue Constructs

The 3D skin constructs were fabricated via aluminum machining, molding,^{[ 51} , ^{52 ]} and double‐casting.^{[ 53 ]} Briefly, as shown in Figure S8a (Supporting Information), a master mold containing a 2‐mm‐diameter hemispherical indentation that was 1 mm deep at the center was first fabricated by lathe machining (HL380, Hwacheon, Changwon, S. Korea; operated by the Instrumentation Development Support Team at Korea Basic Science Institute) with aluminum. Then, the master mold was used to cast an inversed mold (no. 1) in PDMS (Sylgard 184, Dow Corning, Midland, MI). Degassed PDMS prepolymer (prepared with a ratio of 10:1) was poured onto the aluminum mold, followed by curing at 70 °C for 2 h. The inversed PDMS mold was peeled off from the master, then cut into a desired size as a positive mold for the second molding process. The second replica mold (no. 2) was fabricated by silanizing the PDMS master with (tridecafluoro‐1,1,2,2‐tetrahydrooctyl)‐1‐trichlorosilane (CAS# 78560‐45‐9, UCT, PA, USA) that was vaporized in a vacuum chamber overnight to facilitate the subsequent release of PDMS.^{[ 54 ]} To fabricate a thin membrane, degassed PDMS prepolymer was spin‐coated onto a silicon wafer. The thickness of the PDMS layer could be controlled precisely by adjusting the spin‐coating settings.^{[ 55 ]} The membrane was 20 ± 0.1 µm thick, and it was finally attached to the top surface of the 3D construct by O₂ plasma (CUTE‐1MP, Femto Science Inc., Ansan, Rep. of Korea) treatment. Considering the physical properties of the biomaterials including pSG and hDAT, the 3D skin constructs were made as shown in Figure S5b,c (Supporting Information) using molds 1 and 2, respectively. The position of the indentation relative to the skin depth was controlled by changing the positions of molds 1 and 2. The 3D skin constructs made of three different materials had equal dimensions: 14 mm length, 21 mm width, and 5 mm height (14 × 21 × 5 mm) with 2‐mm‐diameter hemispheres. For the hDAT construct, the skin construct was incubated for 30 min at 37.5 °C before injecting AuNRs into the indentation for surface drying. Finally, the AuNR solution was injected through the side of the tissue construct using a syringe (23G, Misosa, CPL Co., Ltd., Ansan, South Korea).

Fabrication of Bioprinted 3D Complex Tissue Construct

To fabricate the 3D complex tissue construct, the developed bioprinting system was used. This system is capable of using six different materials including thermal plastic polymer and hydrogels.^{[ 56 ]} As solubilized hDAT cannot maintain its shape due to its low mechanical property, it is necessary to have a mold until gelation occurs. Polycaprolactone (PCL) was used to print the mold that containing a protruded hemisphere with 2 mm in diameter at the center of lower surface and surrounded by a rectangular wall. Then, hDAT bioink containing MCF‐7 was dispensed at the lower surface at 1 mm thickness and incubated at 37 °C for 30 min. Thereafter, the cell‐free hDAT bioink was dispensed on the MCF‐7 containing layer at a thickness according to the experimental group and incubated at 37 °C for 30 min. After removing the mold, pSG layer was added and the bioprinted 3D complex tissue construct was turned over.

Cell Viability Evaluation within the Bioprinted 3D Complex Tissue Construct

The bioprinted 3D complex tissue construct was transferred to a confocal dish and the viability of cells was evaluated using a live/dead cell assay kit (Lonza, Walkersville, MD, USA). Ethidium homodimer (EthD‐1; 1 µL) and 0.2 µL calcein AM were mixed with 1 mL PBS, added to the constructs, and incubated for 30 min in 5% v/v CO₂ at 37 °C. Live/dead cells were observed using a confocal microscope (IX81, Olympus, Tokyo, Japan). Calcein AM (green) and EthD‐1 (red) represented live/dead cells, respectively. The number of live (green) and dead (red) cells were calculated using a custom‐made macro in NIH ImageJ software (http://rsb.info.nih.gov/ij/plugins/index.html).

Microscopy and Image Analysis

To analyze the heat released by the AuNRs embedded in the 3D tissue constructs upon laser exposure, IR microthermography was introduced for IR thermal analysis. IR microthermography is a useful method for measuring the temperature distribution of a sample surface, because of its high spatial resolution, speed, and non‐contact nature.^{[ 57 ]} An IR microscope system with an IR thermal imaging camera (SC500, FLIR System Inc., USA) was first installed as shown in Figure S1 (Supporting Information). The system had a spectral response in the 3.5–5.1 µm range and a noise equivalent temperature difference of 50 mK. A laser beam (λ = 808 nm, 0.8 mm diameter, 10 W output power from a 400 µm core fiber, FC‐808‐10000‐MM4‐SMA‐1‐1, RGB laser system., Hungary) was expanded to 5 mm diameter by a collimating lens to cover the AuNRs and then was used to penetrate the 3D tissue construct cross‐sections at right angles. The AuNRs located on the 3D tissue construct cross‐sections were heated by the passing laser. Finally, the IR micro thermography system measured the IR thermal radiation of the AuNRs in a hemisphere at the center of each tissue construct, rather than the temperature of the AuNRs. Then, nonlinear transfer functions^{[ 58 ]} were used to convert the IR thermal radiation into a temperature (or a quantitative temperature distribution). The measurement principles in more detail have been explained in previously published papers.^{[ 59 ]} The calculated temperature was confirmed by comparison with the directly measured temperature using a TEC source (2510‐AT, Keithley Instruments, Inc) and RTD sensor (S245PD06, MINCO Products, Inc.).

Statistical Analysis

Data analysis and graphs were performed using Origin software (OriginLab Corporation). All data with error bars were presented as mean ± standard deviation (SD). All experiments comprised of at least three independent experimental batches performed under identical conditions.

Conflict of Interest

The authors declare no conflict of interest.

Authors’ Contributions

Concept and design: J.Y. Bae, K.‐H. Nam. Development of methodology: K.‐H. Nam. Acquisition of data: J.Y. Bae, C.B. Jeong, M. Ahn, S.‐J. Ahn, K.‐H. Nam. Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J.Y. Bae, H. Hur, K.‐H. Nam. Writing, review, and/or revision of the manuscript: J.Y. Bae, HM. Kim, G.H. Kim, Y.‐M. Lim, D.‐W. Cho, K.S. Chang, K.‐H. Nam. Administrative, technical, or material support (i.e., reporting or organizing data, construct databases): C.B. Jeong, D. U. Kim, K.‐S Lee, M. Ahn, J. Jang, D.‐W. Cho, S.‐J. Ahn, H.‐J. Gwon, Y.‐M. Lim, K.‐H. Nam. Study supervision: K.S. Chang, K.‐H. Nam.

Supporting information

Supporting Information

Nam K.‐H., Jeong C. B., Kim H., Ahn M., Ahn S.‐J., Hur H., Kim D. U., Jang J., Gwon H.‐J., Lim Y.‐M., Cho D.‐W., Lee K.‐S., Bae J. Y., Chang K. S., Quantitative Photothermal Characterization with Bioprinted 3D Complex Tissue Constructs for Early‐Stage Breast Cancer Therapy Using Gold Nanorods. Adv. Healthcare Mater. 2021, 10, 2100636. 10.1002/adhm.202100636

Data Availability Statement

Research data are not shared.