Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy

FEASIBILITY EVALUATION OF THE USE OF GOLD NANOPARTICLES AS THERAGNOSTIC AGENTS FOR PERSONALIZED AND PRECISION RADIOTHERAPY José Antonio López Valverde Thesis presented to obtain the degree of Doctor Department of Medical Physiology and Biophysics Medical School of the University of Seville Director Prof. Antonio Leal Plaza MARCH, 2024 II – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy José Antonio López Valverde – III Acknowledgments A doctoral thesis encompasses a great deal. It has led me through numerous experiences that have helped me grow as a person and introduced many new people in my life who have not only been part of it during this period but have also undoubtedly left their mark and will continue to do so forever. Therefore, I want to begin this thesis by expressing my gratitude to all those who have made this work possible. On the institutional side, first and foremost I want to thank the Department of Medical Physiology and Biophysics, which hosted me for the entire duration of the thesis. More specifically, I would like to thank the radiophysics services of the Hospital Universitario Virgen Macarena and Hospital Universitario Virgen del Rocío for providing us with clinical data and for allowing us to use the clinical equipment such as the CT scanner and the linear accelerator. Also, my thanks go to the National Accelerators Center (CNA) and the University of Seville research central services (CITIUS) for facilitating access to the equipment needed for some of the experiments. I want to thank the external labs involved in collaborative efforts crucial for carrying out this work. Many thanks to Ana Paula Zaderenko for welcoming me into her lab and supporting the first stage of gold nanoparticles synthesis. Also, my gratitude goes to the Radiophysics lab at Universidad Santiago de Compostela, particularly to Diego González, for his involvement and support with nanoLIC. Furthermore, this thesis gave me the opportunity to broaden my experience in an international context, where I could taste the flavors of science being more about science and less about pointless administrative and bureaucratic tasks that nothing have to do with science itself. I would like to extend my gratitude to all the people who warmly welcomed me at the Massachusetts General Hospital, Harvard Medical School, and Dana Farber/Brigham and Women’s Hospital affiliated groups. There, I had the opportunity to collaborate on this short-term research stay, which not only taught me valuable scientific techniques applicable to this work but also enriched me personally through an unforgettable experience. Special thanks to Alejandro, without whom this stay wouldn’t have been possible. IV – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy I also want to thank to my peers in the department, and more precisely Maria, as the unofficial reference postdoc I never had, for all the biological methodology. Thanks for all the moments and the time dedicated to solving some of my questions that arose on this side of the project, and for teaching me everything from how you coat your coverslips to how to play ice hockey with discarded dry ice or how to scare my labmates with pressurized popping microcentrifuge tubes. I would especially like to thank everyone from my lab, my academic family, starting with my PhD advisor, Antonio. Thanks for accepting me into your research group from the very beginning, when I was just a dissident student complaining about the lack of training in mathematics, physics, and technological approaches in the Biomedical Sciences bachelor’s program, where I was his student in the biomedical image processing course. It goes without saying that your mentorship, support, and critiques have not only shaped the research work of this thesis but also transformed my vision and approach to academia and life in general. I must also thank Elisa, who tirelessly taught me what Monte Carlo modeling was all about, never considering me the “mudblood” of what this physics lab was before I arrived. Apologies (though not really, sorry not sorry) for introducing the chaos of biology to the world of spherical cows in a vacuum. Many thanks to my peers Ana, Natalia, Carlinhos, Héctor, Inés, Marina, Saravanan… All of you were instrumental in making the time spent on this project not only a fruitful time but also more than enjoyable. Huge thanks to my girlfriend, Emi, for your patience, support, and understanding. From the days when I had to cancel our plans to those when plans shifted to spending nights or holidays together in the lab, playing with spare cells and learning how to culture them while I conducted key experiments. Last and most importantly, a lot of thanks to my family for being so supportive and believing in me throughout this entire journey. You have shown me that every challenge is an opportunity to grow. Your sacrifices, love, and encouragement have inspired and shaped the person I am today. Gracias, Mamá, Papá, Daniel, y por qué no, también Jagger y Tina. Gracias, from the bottom of my heart, for everything. José Antonio López Valverde – V The results and methodology presented in this work have led to the following publications in scientific journals and conference presentations: López-Valverde, J. A., Jiménez-Ortega, E., & Leal, A. (2022). Clinical feasibility study of gold nanoparticles as theragnostic agents for precision radiotherapy. Biomedicines, 10(5), 1214. Skwierawska, D.*, López-Valverde, J. A.*, Balcerzyk, M., & Leal, A. (2022). Clinical viability of boron neutron capture therapy for personalized radiation treatment. Cancers,14(12), 2865. (*Co-first authorship) Jiménez-Ortega, E., González-Castaño, D., López-Valverde, J. A., Miras, H., Gómez, F., & Leal, A. (2022). Physical verification of the dose enhancement by the presence of gold nanoparticles. Physica Medica: European Journal of Medical Physics, 104, S176. López-Valverde, J. A., Jiménez-Ortega, E., Garrido, N., Miras, H., & Leal, A. (2022). Study of gold nanoparticles as theranostic radiosensitizers under clinical conditions. Physica Medica: European Journal of Medical Physics, 104, S123. Jiménez-Ortega, E., González-Castaño, D., López-Valverde, J.A., Gómez, F., & Leal, A. (2022). NanoLIC: A Liquid-Filled Ionization Chamber for Experimental Verification of Dose Enhancement with Gold Nanoparticles. Medical Physics, 49(6), E327-E327. López-Valverde, J. A., Jiménez-Ortega, E., Ureba, A., & Leal, A. (2021). Feasibility study of the use of nanoparticles as theragnostic agent in radiotherapy. Physica Medica: European Journal of Medical Physics, 92, S86-S87. López-Valverde, J. A., Chavarría, I., Jiménez-Ortega, E., Arráns, R., Velazquez, S., & Leal, A. (2021). Gold nanoparticles and SABR for treatment planning of intact breast tumor. Physica Medica: European Journal of Medical Physics, 92, S51-S52. VI – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy José Antonio López Valverde – VII Index Acknowledgements ..................................................................................................................... III Index ................................................................................................................................................. VII Figures index ...................................................................................................................................XI Tables index ................................................................................................................................ XIII Abbreviations ............................................................................................................................... XV Disclosures .................................................................................................................................. XVII Funding .................................................................................................................................... XVII Institutional review board statement .......................................................................... XVII Conflicts of interest ............................................................................................................. XVII 1. Abstract ..................................................................................................................................... 1 2. Introduction ............................................................................................................................ 3 2.1. The current role of radiotherapy in cancer treatment .................................. 3 2.2. Theragnostic approach for adaptive RT and dose painting ........................ 7 2.3. Nanoparticles in radiotherapy as theragnostic agents ................................. 9 2.4. Gold nanoparticles as contrast agents ...............................................................14 2.5. Gold nanoparticles as dose enhancers ...............................................................17 2.6. Gold nanoparticles as radiosensitizers ..............................................................20 2.7. Potential clinical transferability of AuNP in RT .............................................24 2.8. Biological dosimetry .................................................................................................27 3. Hypothesis and objectives ...............................................................................................31 3.1. Hypothesis ....................................................................................................................31 3.2. Objectives ......................................................................................................................31 4. Materials and methods ......................................................................................................33 4.1. Previous methodological considerations .........................................................33 4.2. Gold nanoparticles synthesis and characterization ......................................34 4.3. Gold nanoparticles as contrast agents ...............................................................38 VIII – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy 4.4. Gold nanoparticles as dose enhancers ...............................................................44 4.4.1. Linear accelerator modeling and beam characterization .................44 4.4.2. Gold nanoparticles virtualization, AuNP medium modeling, and simulation settings ...................................................................................................................51 4.4.3. Study of key parameters for dose enhancement ..................................56 4.4.4. Evaluation of dose enhancement in the treatment planning optimization ................................................................................................................................60 4.4.5. Dose enhancement experimental verification ......................................62 4.5. Gold nanoparticles as radiosensitizers ..............................................................64 4.5.1. Cell culture and irradiation...........................................................................64 4.5.2. Immunofluorescence assay for evaluating DNA damage – γH2AX foci……………………………………………………………………………………………………………68 4.5.3. Cytogenetic assay for evaluating DNA damage – micronuclei ........73 4.5.4. Radiosensitization modeling ........................................................................75 4.6. Animal models experimentation – ethics committee proposal ...............79 5. Results and discussion.......................................................................................................81 5.1. Nanoparticle synthesis ............................................................................................81 5.2. Gold nanoparticles as contrast agents ...............................................................87 5.3. Gold nanoparticles as dose enhancers ...............................................................96 5.3.1. Dose enhancement dependence on key parameters ..........................96 5.3.2. Gold nanoparticles as dose enhancers in clinical datasets ............ 105 5.3.3. Dose enhancement experimental verification ................................... 117 5.4. Gold nanoparticles as radiosensitizers ........................................................... 119 5.4.1. External beam radiotherapy ..................................................................... 119 5.4.2. Brachytherapy ................................................................................................ 124 5.4.3. Radiosensitization modeling ..................................................................... 125 5.5. Ethics committee for animal experiments .................................................... 127 José Antonio López Valverde – IX 6. Conclusions ......................................................................................................................... 129 7. References ........................................................................................................................... 131 8. Supplementary material ................................................................................................ 153 8.1. AuNP UV-Vis spectrum analysis code ............................................................. 153 8.2. AuNP TEM images analysis code ...................................................................... 156 8.3. Dose enhancement factor calculation code .................................................. 160 8.4. CN calculation code ................................................................................................ 161 8.5. SoFoCo optimization results for γH2AX ......................................................... 162 8.6. Gold nanoparticles distribution in the radiosensitization simulation 164 8.6.1. Sample script for the distribution of AuNP out of the cell ............. 164 8.6.2. Sample script for the distribution of AuNP inside the cell ............ 165 8.6.3. Number of nanoparticles used for each condition ........................... 166 8.7. Ethics committee for animal experiments proposal ................................. 167 XVI – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy MRI Magnetic resonance imaging MU Monitor unit MV-CBCT Mega-voltage cone beam computerized tomography NTCP Normal tissue complication probability OARs Organs at risk PDD Percent depth dose PET Positron emission tomography PEG Polyethylene glycol PSF Point spread function PTV Planning target volume PVE Partial volume effect ROS Reactive oxygen species RT Radiotherapy r.t. Room temperature SABR Stereotactic ablative radiotherapy SBRT Stereotactic body radiotherapy STED Stimulated emission depletion SUV Standardized uptake volume TEM Transmission electron microscope TCP Tumor control probability TOPAS Tool for particle simulation TPS Treatment planning system VMAT Volumetric modulated arc-therapy Z Atomic number José Antonio López Valverde – XVII Disclosures Funding The author of this work has been supported by the funding from the FPU call from the Spanish Ministry of Science, Innovation, and Universities, reference number FPU19/02950. The experimentation of this work has been possible thanks to the funding from: • Grant RTI2018-098656-B-100 funded by “ERDF A way of making Europe”, by the “European Union”, and MCIN/AEI/10.13039/501100011033. • Project P20_01053 funded by the European Union and the Junta de Andalucía through the European Regional Development Fund (FEDER). • Grant PRYES235123LEA funded by the Scientific Foundation of the Spanish Association Against Cancer. Institutional review board statement Informed consent was obtained from all subjects involved in the study. All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. The studies carried out in this work were approved by the Ethics Committee of the Hospital Universitario Virgen Macarena de Sevilla for the study entitled “Integración de la imagen PET/CT en una aplicación radioterápica de precisión y adaptativa” with reference C.P. DP-PET/CT-C.I. 1958-N-20, and by the Ethics Committee of the Hospital Universitario Virgen del Rocío for the study entitled “Estudio de intervención experimental, (PRETEST-POSTEST) en pacientes con cáncer de mama tratadas con SBRT/SABR estadios I-II (cT1-2cN0, pN1mic-sn M0) o inoperables de inicio, oligometastásicas o con recidiva local”. Conflicts of interest The author declares no conflicts of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the work, or in the decision to present it. XVIII – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy ABSTRACT AND KEYWORDS José Antonio López Valverde – 1 1. Abstract Gold nanoparticles arouse a special interest in radiotherapy. Multiple studies have already shown their potential use as diagnostic and therapeutic agents. In diagnostics they can be used as contrast agents, while in therapy they could act as dose enhancers and radiosensitizers. The purpose of this study is to synthesize and characterize a nanoparticle formulation for this application, to explore the possibilities of gold nanoparticles as theragnostic agents under clinical conditions, to assess the radiosensitization provided in this scenario, and to settle the base scenario for a later phase of preclinical experimentation with animal models. Gold nanoparticles were firstly synthesized with a bottom-up chemical approach and later commercially acquired, and all of them were characterized. They were tested as contrast agents under realistic clinical conditions in both CT scanners and megavoltage cone-beam CT (MV-CBCT) scanners for image-guided radiotherapy. The dose enhancement benefit of these nanoparticles under different conditions and their dependence on critical parameters was evaluated. Furthermore, their therapeutical performance was assessed in optimization treatment planning of actual clinical cases, by virtualizing the presence of nanoparticles in the tumor target considering an efficient targeting. Moreover, gold nanoparticles dose enhancement experimental verification was attempted. Additionally, the radiosensitization effect was evaluated using immunofluorescence and cytogenetic in vitro experimentation, and a model of the associated radiosensitization was proposed using TOPAS and TOPAS-nBio code. Lastly, the bases were settled for future animal experimentation. Gold nanoparticles performed successfully in a theragnostic scenario under the clinical conditions of MV-CBCT incidence beam over not deep tumors such as head and neck and breast cancer cases. A concentration around 20 mg Au/mL inside the tumor volume, in combination with this energy beam, showed a dose enhancement superior to the 15% and radiosensitization of more than 45%. 2 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy INTRODUCTION José Antonio López Valverde – 3 2. Introduction 2.1. The current role of radiotherapy in cancer treatment Cancer remains a leading cause of morbidity and mortality in developed countries. The latest report indicates that, in Spain during 2022 and the first half of 2023, cancer was the primary cause of death among men and the second leading cause of death in women [1]. According to GLOBOCAN, the global incidence of cancer in 2022 was of 20 million new cases, with nearly 10 million fatalities worldwide. Projections for 2040 estimate an increase to 29.5 million new cases and 16.4 million deaths globally [2, 3]. In countries with high gross domestic products, over 50% of cancer patients receive radiotherapy (RT) as part of their treatment regime, either to target the disease locally or to manage and mitigate symptoms of irrecoverable cases, depending on the cancer stage [4]. Breast cancer is the cancer with the highest incidence, even without stratifying by sex, accounting for more than 2.3 million new cases in 2022 alone [2]. The conventional treatment for early-stage breast cancer involves breast-conserving surgery followed by whole breast radiotherapy. Nevertheless, there is growing interest in partial breast irradiation for low-risk patients, which involves delivering a higher dose per fraction directly to the site of the tumor. This approach is under consideration due to the frequent observation of local recurrences near the tumor area and the advancements in the precision of dose delivery now available in radiation therapy centers. As reported by Atun et al. [5], current estimations regarding the global demand for RT by cancer patients indicate that RT is the treatment of choice for 87% new cases of breast cancer patients, according to clinical guidelines and evidence-based medicine, whether as a standalone treatment or in conjunction with other therapies. For head and neck cancer, the percentage is slightly lower at 74%, with a local control rate at five years achieved in 34% of cases, and an overall survival benefit of 20%. In other disease sites, RT is usually required as a prophylactic agent following surgery. Moreover, in locations like the lungs, where RT is utilized in 77% of cases, the accuracy of radiation beams allows using RT under ablative conditions. This is INTRODUCTION 4 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy the case of Stereotactic Ablative Radiotherapy (SABR), the treatment of choice when solid tumors are inoperable. These data point to the increasing importance of RT over surgery and chemotherapy in the treatment of localized solid tumors, the most widespread expression of cancer. Among other reasons, this heightened interest in RT is attributed to technological advances in devices associated with delivering the dose and shaping its relative distribution, adapting the beam to the tumor volume while safeguarding healthy tissues. The use of multi-leaf collimator (MLC) devices has revolutionized treatment delivery through the intensity-modulated radiotherapy (IMRT) technique, employing either static or dynamic methods such as volumetric modulated arctherapy (VMAT). MLCs shape the beam with software-controlled elements made of high atomic number materials for shielding, producing a finely resolved beam with sub-centimeter precision. The success of intracranial radiosurgery has spurred the development of stereotactic body radiotherapy (SBRT) for extracranial lesions. SBRT, a technique born from high-precision targeting, allows for hypofractionation – treatment over fewer sessions. The treatment plan is distributed along time, allowing the cell repair processes to occur in healthy tissues while the tumor cells are subjected to deterministic dose rates. The shift towards fewer sessions is becoming an increasingly used strategy, as it leads to an uptick in biological effective dose, proposing a new paradigm in radiotherapy. In short, new techniques under development are aiming to improve the therapeutic index of radiotherapy, i.e., the balance between the tumor control probability (TCP) and potential complications in the healthy organs at risk (OARs), known as normal tissue complication probability (NTCP). These new RT dose schemas and treatment planning methods, such as SBRT, highlight the need to address uncertainties associated with these new methods [6], which are also related to the individualized biological response of each patient. Cells of the tissues can be damaged directly and indirectly. Whereas direct damage is related to ionization events affecting deoxyribonucleic acid (DNA) and causing alterations such as strand breaks, indirect damage has to do with the creation and imbalance of reactive oxygen species (ROS) inside tumor cells and their INTRODUCTION José Antonio López Valverde – 5 microenvironment, mainly as a product of water radiolysis. As it will be discussed later, while the energy deposition by radiation particles can be precisely quantified and modeled by complex algorithms, accurately assessing the subsequent radiobiological effects remains a significant challenge. The evolution of imaging technology has significantly impacted RT. Medical imaging provides a dual purpose in RT: it aids the specialists in achieving a better diagnosis and serves as the basis for therapy by planning the optimal use of radiation sources and calculating doses in the patient. Hence, medical imaging is crucial for defining the target volumes and disease extension, while also outlining healthy tissues to plan the appropriate treatment while avoiding established toxicity levels of dose in those healthy structures. Imaging in RT is usually limited to threedimensional anatomic information from conventional computerized tomography (CT), and less frequently from magnetic resonance imaging (MRI). Indeed, relying in morphological imaging without accounting for biological/functional considerations is one of the main current limitations in conventional radiotherapy. Image-guided radiotherapy (IGRT) represents an ambitious strategy that incorporates the time as a critical dimension, effectively minimizing geometrical uncertainties related to patient positioning and the dynamic changes in tumor volumes and location throughout the treatment. Generally, the macroscopic tumor volume visible in imaging is outlined as the gross tumor volume (GTV). Subsequently, a clinical target volume (CTV) is often defined, considering the probability of microscopic disease extension. From the CTV, the planning target volume (PTV) is contoured, adding margins around the volumes defined above to account for uncertainties associated with potential morphologic or positional changes in the treatment room, either intrafraction – i.e., during a session – or interfraction – i.e., between sessions of a treatment. In some cases, an internal target volume (ITV) is also delineated, encompassing the CTV and margins for internal movements of the tumor volume, such as those caused by the breathing movement. These margins and volume definitions, based on the morphologic information of the image, are established using clinical-pathological data from surgically removed tumors, adopting a generalized approach grounded in population-based evidence. INTRODUCTION 6 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy Recent advances in imaging guidance technologies and other technological advancements have significantly improved the precision of dose delivery and minimized uncertainties, facilitating the application of techniques such as SBRT and SABR across various disease sites. These methods enable the delivery of highly focused doses to small primary tumors or metastatic areas while protecting adjacent organs. For example, in breast cancer treatment, SBRT/SABR is being explored in both postoperative and, to a lesser extent, preoperative settings. The Hospital Universitario Virgen del Rocío (HUVR) is conducting a study to evaluate the feasibility, safety, and effectiveness of preoperative radiotherapy delivered in a single fraction for selected breast cancer patients. It is now feasible to propose these approaches even in novel scenarios, such as radiation-only breast tumor treatment, a controversial approach described by few authors and adopted in a limited number of clinical trials [7, 8]. Indeed, single-session treatments often challenge the adherence to toxicity dose constraints for organs at risk. Some clinical trials, including the one at HUVR, have explored sophisticated techniques like non-coplanar irradiation angles. In these contexts, the application of radiosensitizers or local dose enhancers could support improved local dosing in breast tumors for SABR treatment, mitigating technical difficulties and addressing some positioning uncertainties by enhancing the dose precisely where the radiosensitizers are located. Despite the technological advancements in hardware and software improving the life expectancy of RT patients, there is a consensus that clinical efficiency may have reached its peak with technology. It is well recognized that patient responses to RT can vary significantly due to tumor heterogeneity and specific genetic differences. Currently, dose prescription and treatment planning predominantly rely on evidence-based population medicine, with limited incorporation of biological considerations in the planning process. This is typically done using parametrized mathematical models to predict the outcome of RT based on dose distribution characteristics [9, 10]. Ling et al. [11] introduced the concept of the biological target volume (BTV), defined as one or more sub-volumes usually included in the GTV, characterized by unique attributes found on functional or molecular imaging techniques. The best way to provide a personalized and precise radiotherapy INTRODUCTION José Antonio López Valverde – 7 treatment that improves clinical outcomes is by including biological information about each patient in the image for diagnosis and planning – i.e., following theragnostic methodology [9]. Although scientific advancements in functional imaging are paving the way for theragnostically guided RT, several challenges remain unresolved. 2.2. Theragnostic approach for adaptive RT and dose painting Image in RT provides an adequate scenario for comprehensive theragnostic approaches that co-develop diagnostics and therapeutic tasks. A theragnostic strategy enables treatment planning based on patient images taken over the sessions to consider the illness’s response to the irradiation. The challenge lies in managing information about variations in microenvironmental or cellular phenotypes that modulate radiation effects. Despite the availability of multiple functional image modalities, such as functional CT using contrasts or dynamic studies, functional MRI, or positron emission tomography (PET), physiological information is rarely considered to parametrize the disease status and evolution in each patient for treatment planning. Dose painting (DP) strategies in RT propose that, to maximize local tumor control, the radiation dose should be ‘painted’ based on the patient-specific tumor biological properties [9, 11], in contrast with conventional RT, which aims to deliver a uniform dose to the whole target volume defined using morphological images. Hence, DP refers to the ability of further customizing the delivered dose distribution based on the pertinent biological information derived from patient imaging. Imagebased dose painting leads to the prescription and delivery of a non-uniform dose to the CTV, relying on biological properties, thus unveiling a novel paradigm in RT, albeit still with a low clinical application as of now. Theoretically, the advent of DP should be unstoppable, not only because it incorporates biological information for detection and target definition in RT using mathematical algorithms such as the one presented by Foster et al. [12], but also because it allows a more personalized response monitoring and facilitates planning changes, if necessary, after the treatment has started. Recent studies carried out by the research group associated with this work demonstrate that incorporating morpho-functional images into INTRODUCTION 14 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy 2.4. Gold nanoparticles as contrast agents Although interest in functional imaging is increasing as essential information to be included in the planning process, as mentioned above, among all image data from devices such as MRI or PET, the image form CT remains necessary for dose calculation based on the electronic density of each material that is part of the tissues. Thus, any approach based on an information-enriched CT image is more appealing than another image modality, which would require some previous conversion process before being directly implemented in TPSs. As already mentioned, CT anatomical information is clearly not enough for personalized RT, but the idea of turning the well-known CT device already installed in any hospital into a functional imaging device attracts a great deal of attention. Among several nanoplatforms proposed for RT, gold nanoparticles (AuNP) are one of the most extensively studied due to their high X-ray absorption coefficient (Z=79), ease of synthesis, and biocompatibility, resulting in a lower toxicity. They exhibit a longer vascular retention time compared to other contrast agents as a result of their higher molecular weight and have been shown to passively accumulate in tumors while remaining outside healthy tissue when their longest dimension is less than 100 nm [34, 37, 69, 70]. In this way, AuNP represent a theragnostic research area of high interest in oncology [36, 71]. Nevertheless, the functional feature of an efficient nanoplatform could require both passive and active targeting to achieve the necessary mass concentration in the tumor site. Despite the variety of AuNP formulations, in most of them, two parts can be distinguished: the core of the nanoparticle, primarily made of gold, and the shell or surface, composed of organic molecules that functionalize the nanoparticle. The shell is the key feature to consider for the cell-nanoparticle interaction, as the surface molecules are responsible of the nanoparticles’ targeting, stability, immunogenicity, and toxicity, among others, while the core plays a crucial role in radiosensitization. It was previously thought that some nanoparticles could perpetually remain intact in some tissues. However, Balfourier et al. reported that small AuNP were indeed degraded in vitro by cells through lysosomes, forming aurosome-like structures [72]. INTRODUCTION José Antonio López Valverde – 15 Gold exhibits a higher X-ray attenuation coefficient than iodine (Z=53) or barium (Z=56), both of which are frequently used as contrast agents in radiodiagnostics. The role of efficient functionalization could become essential for achieving the higher sensitivity or contrast enough for diagnostic purpose. Fortunately, AuNP offer numerous opportunities to find the optimal formulation for targeted delivery, colloidal stability, and blood retention necessary for a timeoptimal imaging and treatment delivery processes [73]. Among studies focused on assessing parameters involved in biodistribution and targeting, it seems that some consensus range of values has been achieved for some of them, such as the size of AuNP, which should be as large as possible to maximize mass concentration and smaller than the diameter of the vascular pores (100 nm) for passive or active tumor targeting [74]. Conversely, larger nanoparticles exhibit slower maximum accumulation (8 hours after injection for 100 nm particles) and a gradual decrease after 24 hours, unlike the continuous increase observed for smaller particles (15 nm) [75]. Applying polyethylene glycol (PEG) onto the AuNP surfaces, one of the most widely studied surface functionalizations, appears to improve colloidal stability, thereby extending the blood half-life sufficient for achieving the necessary mass concentration for contrast. These durations are longer compared to other conventional contrasts agents like iodine, which has a half-life of minutes, allowing the same nanoparticle injection to potentially facilitate subsequent radiation sensitization during treatment if planning based on the patient image with AuNP can be solved within this time interval. As commented, the use of AuNP as contrast agents has been extensively studied. As low X-ray energies are used – i.e., around the hundred keV – the interaction with the higher probability is the photoelectric effect, as observed in Figure 2. The probability of interaction depends on Z to the third power according to Braggs- Pierce law: 𝜏 = 𝐾 · 𝑍3· 𝜆3. Therefore, changes in Z result in a natural contrast in the image caused by the X-ray attenuation. Thus, gold serves as an excellent contrast agent for CT and low X-ray based techniques, especially within the human body, where the predominant elements, such as H, C, N, and O, have low atomic numbers. INTRODUCTION 16 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy Figure 2. Relationship between the incident photon energy and the atomic number (Z) highlighting the most probable radiation interaction effect produced with the matter. The horizontal line represents the gold atomic number (Z=79). Adapted from López, C [76]. On the other hand, parameters regarding the CT scanner such as tube voltage or even processes such as software reconstructions influence the image. A standard protocol may be necessary to obtain the contrasted image if AuNP mass concentration is not too high. In this sense, the new dual-energy CT devices provide information about how substances behave at different energies, since the closer their energy level is to the K edge of a substance, the more the substance attenuates the radiation. Understanding a substance’s behavior at two different energies facilitates the use of AuNP as contrast agents for functional imaging [73]. Consequently, target can be performed more accurately on post-processed images for a DP approach in the treatment planning. These kinds of approaches could enhance CT sensitivity as morpho-functional imaging device, compared to other clinical imaging modalities like PET, while also leveraging its inherently higher spatial and temporal resolution. However, for lower concentrations of AuNP, distinguishing neighboring soft tissues might require increasing the radiation dose beyond that used in conventional CT studies, potentially compromising the contrast level with patient safety. At this point, it seems clear that a coherent methodology should seek to find the minimum mass concentration able to provide contrast in the CT image with the lowest radiation dose possible. Subsequently, it should attempt to control the INTRODUCTION José Antonio López Valverde – 17 biological parameters, such as AuNP biodistribution and tumor uptake, by using an injectable concentration that does not induce acute or long-term toxicity. 2.5. Gold nanoparticles as dose enhancers In contrast to the diagnostic application, the therapeutic use of gold nanoparticles is less studied. Therefore, the theragnostic use of AuNP remains a novel approach for RT planning. Regarding the therapeutic efficacy of cancer treatment employing radiation sensitization with nanoparticles, another important aspect beyond the features of nanoparticles must be considered: the radiation beam for RT treatment. The idea is based on the enhancement of electron release in the medium due to the interaction of the incident beam with the high-Z element at the tumor site. Among iodine, platinum, and others, AuNP have again been the nanoparticles most intensively studied as radio-enhancers both in vitro and in vivo experiments. An exception to this could be the aforementioned AGuIX and NBTXR3 particles. Several reviews [37, 38, 52] have collected enough published works showing AuNP’s dose enhancement effect under different beam modalities, generally expressing the potential efficiency as a percentage of the dose delivered in the same scenario but without nanoparticles. Radiation modality plays an essential role because the results could be substantially different depending on the incident particles – photons, electrons, ions, or protons. Energy value is also a critical factor to consider [36, 38]. Electron beams are used in external RT for superficial tumors, and the biological effect is not expected to be different to the one obtained with photon beams. On the other hand, proton beams interacting with Au produce secondary electrons from outer atomic orbitals, with a small angle of scattering. Thus, the expected dose enhancement could be lower at larger distances, in the order of micrometers, compared with MeV and keV photons in the range of nanometers [77, 78]. Currently, MeV photon beams are the most desirable for dose enhancement, as the technology and software in RT are better developed for these beams, given their frequent use in clinical applications. Unlike the kilovoltage photon beam used for CT imaging, mainly based on the photoelectric effect, the more widely used therapeutic beam in external RT is the INTRODUCTION 18 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy photon beam in the megavoltage range, where the Compton effect is the dominant interaction (see Figure 2). Energy values higher than 1 MeV can provide an adequate depth dose but have a low interaction with AuNP in the Compton range. Lower energies provide a higher probability of photoelectric absorption and so, a higher interaction with gold in cells, but the dose enhancement might not compensate for loss of penetration through tissues. Despite this, keV beams have been recurrently used in preclinical studies with relatively small animal models, as seen elsewhere [79–86], producing results with a difficult translation to the clinical reality. Unfortunately, the Compton scattering cross-section’s dependency on Z is almost linear, in contrast to the potential relationship with the photoelectric cross-section for keV photons. This could lead to a significant drop in radiation sensitization for these models with small animals when realistic clinical conditions are applied, making them less suitable in this context. Physiologically, the model may be suitable, but it fails to meet the necessary physical considerations. The size of the animal is a determinant for the tumor depth, and hence for the irradiation beam characteristics. On the other side, a significant percentage of MeV photons transform into keV photons and electrons when interacting with beam modifiers, such as MLCs for intensity modulation, and upon penetrating the patient. Therefore, it would be interesting to explore AuNP possibilities under the actual conditions of IMRT treatments. Other studies have demonstrated a superior dose enhancement ratio with AuNP using unflattened beams in linear accelerators labelled as ‘flattening filter-free’ (FFF) [36, 87]. This technique is commonly used in RT for high dose-rate treatments, which is anticipated to be the standard approach in the near future and is already in use for certain current scenarios. For instance, immediately after particles exit the accelerator under predefined conditions, 0.5% of them have energies below 150 keV. In contrast, at a depth of 10 centimeters inside the patient, this fraction increases to 13%, or to 20% if the linear accelerator is FFF [37]. Treating patients with plain keV photons to achieve maximum radioenhancement efficiency is not possible. Such treatments would require excessively long durations compared to current standards, and given the dose buildup characteristic of the beam, the percent depth dose would be unsuitable for effective treatment. High-energy beams deposit the maximum dose beyond the surface, creating a buildup region that extends from the surface to the point of maximum INTRODUCTION José Antonio López Valverde – 19 dose delivery within the patient for that beam incidence. This effect occurs because Compton events predominate over photoelectric events, typically for energy beams exceeding ~1 MeV, given the patient’s elemental composition. Thanks to the buildup effect, it is possible to minimize the dose to the skin and reduce adverse effects, such as dermatitis by combining beams of various incidences and angles and maximize the dose precisely targeted to the tumor mass. Low-energy beams, in the order of keV, deposit their maximum dose in the skin, complicating treatment planning and increasing the likelihood of adverse effects in superficial OARs including the skin. A similar issue arises with electron beams, which, due to their charged nature, exhibit little to no buildup region even at high energies. Consequently, they are utilized for treating shallow tumors. The paragraphs above show how important the implementation of beam interaction modeling in the methodology for the study of AuNP as theragnostic agents in this first physical stage should be. Many works have focused on modeling studies using a numerical method based on Monte Carlo (MC) simulations [38], where the characterization of the beam modality is essential. Monte Carlo simulations, which generate random numbers to sample well-known probability density functions, determine the outcomes of particle interactions with different atoms in the mediums. This is the gold standard procedure for dose calculation in RT as it has excellent experimental support at these energies, and nowadays, computational resources permit a high sampling magnitude, reducing the statistical error below 1%. The recurrent use of MC simulations in RT for dose calculations in complex cases, particularly under special RT techniques, reflects their critical role and reliability, because the analytic algorithms implemented in TPSs are considered only approximations based on standardized experimental measurements [88, 89]. It is important to note here that although some radiobiological models have been implemented in MC simulations, to date, these MC models can simulate physical and physicochemical processes of radiation interactions with high precision, yet they fall short of accurately simulating biological processes. Several computational models aim to account for the DNA damage induced by radiation at the physical and chemical stage. The main one is Geant4-DNA code [90], a user code that has been INTRODUCTION 20 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy adapted to be the basis of TOPAS-nBio [91], an extension of TOPAS [92, 93], a toolkit based on Geant4 MC code. Great efforts are being made to define models that not only consider the physicochemical stage of radiobiology, but also consider the biological processes. However, accurately describing these processes through probability distributions, a fundamental requirement for numerical methods, remain challenging. This is why the biological phenomenon is not suitable for a statistic description and the necessary experimental support is tightly related to complex in vitro/in vivo experiments. Thus, while MC modeling can calculate the dose enhancement effect, it cannot consider the radiosensitization effect, and they are not the same thing, as will be discussed in section 2.6 below. 2.6. Gold nanoparticles as radiosensitizers At this point, it is necessary to distinguish between the concept of radiosensitization, specified as the ability to sensitize cells to radiation, and the radiation dose enhancement of AuNP, referring to the increase of the radiation effect in the initial physical process in terms of the sequential events in the evolution of the radiation damage. Unlike radiosensitization, dose enhancement can be easily simulated and parameterized to be implemented in the optimization of the treatment planning process. Nonetheless, dose enhancement does not predict the biological outcome, which depends on many factors beyond the energy delivered at a specific location, even though extra energy is deposited because of the interactions with high Z elements at the site. Some experiments based on keV photon beams exhibited important dose enhancements leading to much lower radiosensitization values than expected. In contrast, higher levels of sensitization were shown in several studies using MV X-rays and electrons, for which the dose enhancement was clearly lower than the one obtained with lower energy beams [94]. The latter explains that radiosensitization depends on factors other than strong photoelectric absorption, such as the increased Auger emission occurring from deexcitations of irradiated Au atoms. These Auger cascades, emitting no more than 10 electrons with energies below a few keV, cannot significantly increase the absorbed dose in the physical stage but could be very important for enhancing radiosensitization in the subsequent chemical and biological stages. On the other hand, the production of ROS INTRODUCTION José Antonio López Valverde – 21 such as hydroxyl radicals, superoxide, and hydrogen peroxide produced from the radiolysis of water can cause damage to cellular DNA or can lead to oxidation of lipids and proteins to generate apoptosis and even necrosis of the tumor cell [95]. Taggart et al. found that mitochondria could play a key role in the radiosensitization mechanism of AuNP [83]. Several in vitro studies showed significant levels of ROS depending on the size, shape, and chemical surfaces of AuNP, and even indicated time-dependent results, but the mechanism involved in oxidative stress by AuNP remains unclear [94]. To assess the radiosensitization effect, DNA damage and oxidative stress are the main mechanisms involved in a possible higher radiobiological effect. Figure 3 shows a diagram of the effects of AuNP as radiosensitizers. Figure 3. Diagram of some of the mechanisms for the radiosensitization promoted by AuNP due to prompt electrons, mainly Auger electrons. Figure created with BioRender. Some works have shown that the highest radiosensitization is obtained with AuNP coated with the shortest possible linker to the cell’s nucleus [96]. These studies are mainly carried out under in vitro experiments and, even when the conditions about the beam setup – i.e., beam modality and energy are fixed, they are very susceptible to small changes in surface properties of the analyzed nanoparticles. Although this type of study is far from clinical application, it INTRODUCTION 22 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy highlights that the binding linkers in the coating can absorb secondary electrons with low energies that will not reach the DNA or water molecules, hence not increasing direct or indirect damage to the cell. The main interaction result expected to improve the radiosensitization is the deexcitation after ionization and excitation of atoms in AuNP due to the interaction with radiation. Some of these electrons are absorbed inside the AuNP, and the number of those which escape (usually no more than 7 electrons) depends on physicochemical characteristics of the nanoplatforms. The free path range of these electrons is not dependent on the beam modality or incidence energy, but on the energy associated with electron transition of the specific atom composing the core of the nanoplatform, so for all situations, most of them fall off within the first 250 nm from the gold nanoparticle. This determines how significant the cell and tumor uptake of nanoparticles is in the expected process. Unfortunately, these biological processes are challenging to measure in in vitro experiments [97], and they cannot be analyzed using in vivo studies, necessary for clinical transfer to assess biological parameters such as toxicity or biodistribution. The first in vivo study showing AuNP as radio-sensitizers was demonstrated by Hainfeld et al. [86] by using 250 keV X-rays on mice with subcutaneous EMT6 mammary carcinomas, obtaining concentrations of 7 mg Au/g of tumor and showing a 1-year overall survival of 86% when AuNP and X-ray were used, vs 20% when only X-ray were used, and 0% when only gold was used. Almost ten years later, Hainfeld et al. [85] performed an in vivo experiment for a theragnostic study on mice with malignant brain tumor where they observed enhanced contrast in micro-CT image and RT enhancement effects with 11 nm diameter AuNP. They measured concentrations up to 15 mg Au/g of tumor, with a ratio of 19:1 on concentrations in the tumor vs healthy tissue. They estimated a dose enhancement of around 300% by using 100 keV X-rays. These and other studies are hopeful, but the clinical transfer requires CT imaging with the resolution for humans, unfortunately far away from the image data provided by micro-CT for animals, as well as radioenhancement at treatment-feasible energies, significantly different to the keV X-rays used in animals, as it has been discussed above. INTRODUCTION José Antonio López Valverde – 23 Several studies have been published on CT images of mice with tumor xenografts injected with AuNP of varying properties, irradiated with X-rays of different energies [79, 98]. Specifically, McQuade et al. [98] found that the mass concentration required for CT contrast (in the order of mM concentrations, grams Au/kg body weight) would require to be three orders of magnitude higher than the concentration for radio-enhancement (μM), so they decided to turn and study MRI alternatives, what was followed by other researchers until the present day. Nevertheless, these considerations should be reviewed for the clinical conditions applied to patients, since dosages should be reconsidered in light of the differential nanomaterial pharmacokinetics compared to conventional drugs and considering that higher concentrations were found nontoxic [85, 86]. Also, as the beam energy increases, so does Compton scattering dominance, resulting in lower dose enhancement for an identical AuNP concentration. This consideration, coupled with the potential higher radiosensitization with MeV photons, encourages carrying out studies using similar mass concentrations for both diagnostic and therapeutic purposes. Although many works focusing on passive targeting have been published, a potential transfer to the clinic may require active targeting approaches, as commented above. RT has a dual objective: delivering the prescribed dose to the PTV while safeguarding OARs from receiving toxic dose levels, which vary for each organ or tissue. This way, the efficiency of targeting is essential, and passive targeting by itself could not be enough, even though there are some promising results such as the ones already mentioned by Hainfeld et al. [85] where a ratio of 19:1 was obtained for tumor vs healthy tissue. Additionally, promoting AuNP- nucleus co-localization is correlated with higher radiosensitization, and this does not seem easy to achieve utilizing passive targeting. In any case, the efficiency linked to using active versus passive targeting will depend on the tumor type and the time elapsed from injection to irradiation. Thus, a universal solution for functionalizing AuNP to amplify the radiobiological effect in all tumor types is not expected. Active targeting can be obtained by bio-conjugating the surface of AuNP with peptides or antibodies against proteins expressed by cancer cells. Hainfeld et al. [35] demonstrated that nanoparticles functionalized with PEG and bio-conjugated with INTRODUCTION 30 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy HYPOTHESIS AND OBJECTIVES José Antonio López Valverde – 31 3. Hypothesis and objectives 3.1. Hypothesis The targeted delivery of optimal concentrations of gold nanoparticles (AuNP) to tumor tissues represents a significant theragnostic advancement in precision radiotherapy, in combination with the adequate incident beam. This enhanced treatment efficacy can be achieved in two different areas thanks to the high atomic number of AuNP. Firstly, they provide morpho-functional contrast in computed tomography (CT) imaging, leveraging CT's high spatial resolution and cost-effectiveness for precise tumor localization and monitoring. Secondly, these nanoparticles act as dose enhancers and radiosensitizers. Moreover, AuNP can be used in advanced IGRT with the adequate incident beam to act as contrast agents and also for therapeutic dose enhancement purpose. 3.2. Objectives The main objective of the work is to determine the possibilities of implementing gold nanoparticles in radiotherapy, exploring the role and performance of AuNP under realistic clinical conditions in diagnostics and therapeutics, and evaluating the radiobiological benefits of these treatments. For achieving this objective, several specific objectives were defined: • Synthesizing and characterizing gold nanoparticles with a robust reproducible protocol, yielding particles stable enough for them to last while the whole treatment planning process is carried out. • Determining the diagnostic capabilities of gold nanoparticles under different clinical conditions using computerized tomography and megavoltage cone-beam equipment. • Calculating the influence of critical parameters in dose enhancement, such as nominal energy of the incident beam, field size, or tumor depth. • Evaluating the role of gold nanoparticles as dose enhancers and their effect in the relative dose distribution optimization process in simulations of actual clinical cases. HYPOTHESIS AND OBJECTIVES 32 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy • Assessing the radiobiological response in terms of radiosensitization under these clinical conditions. • Establishing a starting point for the simulation of radiosensitization in these types of experiments, using a multi-scale model approach. • Settling the basis of the dose enhancement experimental verification procedures and designing further biological experimentation needed implying animal models, to experimentally demonstrate the feasibility and benefit of using gold nanoparticles in the clinic. MATERIALS AND METHODS José Antonio López Valverde – 33 4. Materials and methods 4.1. Previous methodological considerations For designing a robust methodology, it is crucial to consider some essential biological aspects. The first one is that to obtain a contrasted CT image, a specific mass concentration of AuNP is required at the tumoral site, at least. The radiological density obtained should be between soft tissue and muscle, but achieving a higher contrast level, such as one closer to that of less dense bones, is also considered. Anyway, injecting a higher quantity of nanoparticles than what is needed to achieve that target concentration will be necessary, as not all nanoparticles will reach the cancerous tissue: macrophages will capture some of them, others will be metabolized, excreted, or even deposited with low affinity in healthy tissue. A procedural approach could consist of handling different concentrations until a relative contrast level is achieved, and hence, assume that higher concentrations could yield better contrast levels but would be harder to achieve from a pharmacological point of view. From now on, this concentration will be referred to as the radiological concentration. Besides, as stated in the introduction section, a higher concentration may be necessary for achieving a therapeutic effect. This concentration, which will be referred to as the therapeutic concentration, will be higher than radiological concentration because the treatment energies are higher than the energies used in diagnostics, and therefore, there is a higher Compton scattering contribution in therapeutics rather than photoelectric. Finally, it is necessary to consider the existence of a toxicity threshold in nontarget organs such as liver, kidneys, or spleen, limiting the potential concentration values to administrate. As a subsequent step, under optimal conditions and with proper AuNP bioconjugation and functionalization, achieving the radiological and therapeutic concentrations in the target could be highly efficient, allowing for lower quantities to be injected in the patient while the target concentrations are reached. MATERIALS AND METHODS 34 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy It is essential to distinguish between the injected concentrations and concentrations reached in the tumoral tissue, as injected concentrations are expected to be relatively low. Considering classical pharmacology, it is logical to assume that achieving a specific concentration in the tumor mass necessitates a higher concentration in the blood. However, nanoplatforms pharmacology is different from classic pharmacokinetics. With proper functionalization, and even more so with bio-conjugation, the concentration achieved in tumoral tissues should rise above the concentration in blood or in the original injection. Clearly, as presented above, determining such toxicity concentration threshold belongs to the biological aspects of the whole methodology, and it should only be undertaken after the physical experimentation provides us the concentration values previously outlined. The radiological concentration might not require a high biological efficiency, whereas therapeutic concentration might, what would give the radiological image a more functional image value. The final decisions will depend on the toxicity threshold, somehow described for some animal models [106, 107]. All these considerations led us to propose an inverse alternative to what most of the works mentioned in the introduction section proposed. Instead of directly trying to find which concentrations works as contrast agent and/or dose enhancers using in vivo studies, we decided to study the potential benefits in RT planning by using physical models for a given AuNP concentration, assuming it can reach the cancerous area after overcoming the biological considerations outlined above. As will be discussed later, apart from the expected benefit of achieving the scientific objective, this sequential integrated strategy ethically adheres to the three R’s rule for experimental animal studies. Those results will lead to observing the feasibility of the theragnostic approach using AuNP in RT, and if it is feasible, this theragnostic concentration will be defined and demonstrated by further experimentation based on in vitro and in vivo studies. 4.2. Gold nanoparticles synthesis and characterization Two primary sources were used for obtaining the AuNP necessary for the experimental work presented in this thesis: commercially available nanoparticles and nanoparticles synthesized specifically for this project. For the latter, a significant collaboration was established with the research group "Condensed MATERIALS AND METHODS José Antonio López Valverde – 35 Phases and Interfaces Physical Chemistry", led by Prof. Ana Paula Zaderenko Partida at the University Pablo de Olavide. This partnership was contributory in the custom synthesis of AuNP tailored to the specific requirements of this study. The synthesis task was conducted directly by the author of this work, leveraging the expertise and facilities provided by Prof. Zaderenko's group. This collaborative effort ensured that the nanoparticles were synthesized with precise control over their properties, crucial for the subsequent experimental phases detailed herein. The synthesis methodology used was adapted from Dong et al. [108], and consists of a modification of the Turkevich method, known since the early 40s and 50s [109, 110]. This method implies the reduction of chloroauric acid (HClAu4) using sodium citrate as both the reducing agent and the stabilizing agent. Upon heating, the citrate reduces the gold ions to gold atoms, which nucleate and grow into nanoparticles. The citrate ions also adsorb onto the surface of the nanoparticles, preventing agglomeration. The resulting citrate-coated nanoparticles can undergo ligand exchange in an ulterior phase of the process, where PEG can be adsorbed to the surface of the particles. The Turkevich method was chosen because it is the most suitable of the wellknown methods for yielding spherical nanoparticles with diameters in the order of tens of nanometers. This method is highly scalable and allows adjusting the yielded AuNP diameter based on the gold-citrate ratio. The detailed procedures for the synthesis and characterization follow: - Before starting the synthesis, all the glassware that is going to be used should be thoroughly cleaned and rinsed using laboratory-grade detergent capable of removing any possible metallic contaminants effectively. This is fundamental, since it will help avoid any cross-contamination or interference that could act as nucleation sites for nanoparticle formation, leading to unintended sizes, shapes, or compositions of the particles being synthesized. Skipping this step could significantly affect the reproducibility and quality of the AuNP synthesis. - Once the material is ready, the reagent solutions are prepared. The absolute amounts can be scaled as needed [108]. Here, sample amounts are shown for a reaction scaled for taking place in a 100 mL Erlenmeyer flask. A solution of chloroauric acid is prepared by adding 9.8 μL of a chloroauric acid trihydrate MATERIALS AND METHODS 36 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy [Sigma-Aldrich, St. Louis, MO, USA, ref: 520918-1G] stock aqueous solution at 50% w/v to 50 mL of distilled water. On the other side, 10.3 mg trisodium citrate dihydrate [Panreac, Barcelona, ESP, ref: 131655.1211] is diluted in 1 mL of distilled water. Hence, the molar ratio of sodium citrate to chloroauric acid is 2.8. - The chloroauric acid solution (50 mL) is heated to boiling point under vigorous stirring in an Erlenmeyer flask covered with a watch glass. The heating plate is set to 180 °C and the magnetic stirring is set to 900 rpm. When the watch glass crackles, the sodium citrate solution (1 mL) is added all at once. Color changes occur over 5 minutes (see Figure 4). The mixture, initially crystal clear, shifts to a dark purple color in 1 minute, and at two minutes it starts to shift to red. - After 6 minutes, the Erlenmeyer flask is removed from the heating plate and is left for cooling at room temperature (r.t.) on top of a thermal insulation material. The result is 51 mL of a AuNP solution at a low concentration. Figure 4. Visual representation of color change over time during the Turkevich synthesis of gold nanoparticles, illustrating the progression from the initial solution to the final colloidal gold dispersion. Each image captures the reaction mixture in an Erlenmeyer flask at a specific time stamp since the addition of the citrate reagent, shown below each picture. The characteristic transition from almost transparent to deep red, indicative of nanoparticle formation, is highlighted. MATERIALS AND METHODS José Antonio López Valverde – 37 The resulting AuNP solutions were first characterized by measuring their UV-Vis absorbance spectrum using an Ocean Optic spectrometer equipped with an HR4000 detector, with integration times in the order of 3.88·10-3 s, at least 100 scans to average each measurement, and the electric dark correction enabled. This allowed recording spectra ranging from ~195 nm to ~1110 nm with a high resolution (3648 bins). The spectra were analyzed using MATLAB R2020b [Natick, MA, USA] to characterize the surface plasmon resonance peak position, its height in terms of optical density, and the data deviation around the peak. These parameters are related to the average size of the nanoparticles, the AuNP concentration, and the size distribution deviation around the average size, respectively. While the first two measurements are easy to parametrize and measure, the deviation of the UV-Vis absorption distribution, a non-pure Gaussian distribution, can be measured in different ways. We chose to measure it as described in equation 1 below: 𝐷𝑖𝑠𝑝𝑋=𝜆𝑥2− 𝜆𝑥1 𝜆𝑀𝐴𝑋 {1} where x represents a percentage of the maximum absorption wavelength (𝜆𝑀𝐴𝑋), and the variables 𝜆𝑥1 and 𝜆𝑥2 are the wavelengths that intersect that percentage of the maximum absorption. To reduce the noise in the measurements and its impact in the normalization process, the data around the plasmon resonance peak in the interval [450, 600] nm were fitted to a 4-term Gaussian fit. That fitting function was then used for the normalization and calculation of the parameters already mentioned, hence avoiding normalization based on a point potentially produced by a noise spike. The MATLAB script used for characterization of the UV-Vis spectra can be found in the supplementary material section 8.1. Dynamic light scattering (DLS) measurements were taken to get an idea of the hydrodynamic diameters and the size distribution of the nanoparticles, which was further confirmed by transmission electron microscopy (TEM). The TEM micrographs were also compared to the commercial 15 nm core diameter AuNP solution AuroVist™ [Nanoprobes, Yaphank, NY, USA, ref: 1115A-5X40MGAu]. TEM images were analyzed using MATLAB image processing toolbox to automatically segment and identify the nanoparticles, hence obtaining the size distribution of the nanoparticles. Image processing artifacts were filtered out from MATERIALS AND METHODS 38 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy the automatic segmentation process and calculation of the distribution by statistically filtering outliers, defined as data out of the range comprised by 𝜇 ± 3𝜎. Furthermore, the normality of the data was tested using the Shapiro-Wilk test. The MATLAB code script used can be found in the supplementary material section 8.2. The nanoparticles were concentrated mainly by centrifugation at 20 °C and 12000 rpm for 10 minutes without using a brake at the end. Using a rotary evaporator as an alternative for reconcentration after synthesis was also explored as an option. Ligand exchange with PEG-20000 [Sigma-Aldrich, Burlington, MA, USA, ref:81300-1KG] was used at different concentrations (0.5%, 0.25%, and 0.125% w/v) to add colloidal stability and decrease the probabilities of aggregation and precipitation during the reconcentration. The final concentrations of the original synthesis were approximated by measuring in triplicates the dry weight after evaporation of a known volume of particles dispersion, and they were further explored by inductively coupled plasma (ICP) optical/atomic emission spectroscopy, also known as ICP-OES or ICP-AES. The synthesis relative efficiency was calculated based on the amount of gold detected in AuNP relative to the amount of gold originally present in the synthesis. The nanoparticles solutions were stored in the fridge at 4 °C. Their stability was tested over three weeks by observing changes in the UV-Vis absorption spectrum to ensure there was enough time for using the nanoparticles for the methodology of the project from the time that they were synthesized. Additionally, AuNP stability at r.t. was tested for up to five days to ensure that possible transportation temperature spikes would not compromise their stability. 4.3. Gold nanoparticles as contrast agents First, the radiological concentration necessary to offer a distinguishable signal in a clinical CT image was studied. For this purpose, several nanoparticles were used, including newly synthesized nanoparticles, such as the ones described in section 4.2, and two commercial solutions: AuroVist™ 15 nm and Merck commercial 50 nm core diameter AuNP functionalized with amine-group ending PEG-3000 [Sigma-Aldrich, Markham, ON, CAN, ref: 765406-1ML]. The concentration of each AuNP solution was known beforehand and choosing between the different options was dependent on factors such as the availability of the target concentrations for the CT scan. MATERIALS AND METHODS José Antonio López Valverde – 39 In the beginning, serial dilutions of the nanoparticles were prepared and tested in different sets. The aim of this initial approach was to explore the range where the radiological concentration could be, considering clinical conditions. For instance, the lowest concentration array of AuNP concentrations, depicted in Figure 5A, consisted of 7 samples of 0.25 mL prepared in strips of eight 0.2 mL PCR tubes with tethered caps. The first sample in the set represented a concentration of 0.223 mg Au/mL of solution and the rest of the samples show a half dilution relative to the previous one, resulting in the following dilution factors: 1/2, 1/4, 1/8, 1/16, 1/32, and 1/64. Lastly, as a control, the eighth tube was filled with water. Another set of dilutions, shown in Figure 5B, started at a concentration of ~4 mg Au/mL of solution, followed by two dilutions with factors of 1/2 and 1/4, and a control tube filled with water. The rest of the tubes for this set were left empty, with air. These series of wells were used to perform an initial CT calibration and observe the level of contrast provided by each nanoparticle concentration. For this purpose, a CIRS 062 Electron Density Phantom was used to simulate the context of AuNP inside the patient and to compare with other inserts mimicking the physical densities of various tissues: lung (in inhalation), lung (in exhalation), adipose tissue, breast tissue (50% adipose, 50% gland), muscle, liver, trabecular bone (200 mg/cm3), and dense bone (800 mg/cm3). The data on the densities and electronic densities of these standardized inserts are shown in Table 1. To properly align the tubes containing the AuNP and insert them correctly into the phantom, due to their small volume compared to the size of the inserts, the tubes set was inserted into a 50 mL centrifuge tube. This tube had been previously filled with Transonic Gel ultrasound gel [TELIC, Bigues i Riells, ESP, ref: TOG25], which has an electronic density similar to that of water. This ensured that once the wells were inserted, they were fixed in place within the centrifuge tube (see Figure 5D). This tube was placed inside the CIRS 062 density phantom, as a standard insert (see Figure 5E). With this arrangement in the phantom, whose physical density simulates that of water, it was possible to obtain the level of contrast that the nanoparticles would offer in a clinical situation, that is, inside the patient itself, if the proposed concentrations were reached. MATERIALS AND METHODS 46 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy Figure 8. Siemens ONCOR™ linear accelerator target component pictures, courtesy of Juan Hevilla Molina from Siemens. (A) The target rail placed in the accelerator’s head is shown, highlighting the orientation and direction of the electrons from the waveguide; (B) detailed view from the rail where the carbon part of the target can be seen, and this is the part that is closer to the waveguide; (C) detail of the other side of the rail, the one that is farthest from the waveguide, and the aluminum part of the target can be seen. The produced photons orientation is depicted. Both beams had been properly modeled according to the provided technical specifications. The PDDs and profiles generated using MC simulations were plotted and compared with the experimental data, and this comparison achieved high agreement for both accelerators. The Siemens ONCOR™ MV-CBCT beam is only used for IGRT purposes, so the beam characterization carried out by the HUVM radiophysicists is simpler than the standard beam quality control and quality assurance protocols. Unfortunately, the only available experimental data are a PDD with a 10×10 cm2 field size. However, a high agreement with that experimental PDD and its MC simulation, as shown in Figure 9, suggests a credible level of accuracy in our modeling. This confidence is reinforced by the previous successful simulation of the linear accelerator’s key components, such as the MLC. As already mentioned, apart from the removal of the flattening filter and the changes in the target for the MV-CBCT beam, the rest of the components remain consistent with the rest of the beams of the accelerator which was largely tested and validated by the Medical Physics research group at the University of Seville. In short, it is reasonable to assert that if the model accurately predicts the PDD, which is fundamentally influenced by the energy spectrum and also by the spatial distribution of the beam, then the MC model can be trusted to provide an illustrative representation of the characteristics of the MV-CBCT beam, MATERIALS AND METHODS José Antonio López Valverde – 47 even in the absence of extensive experimental data. However, for in vitro experimentation, these uncertainties in the characterization should be considered and mitigated, as it will be discussed in the section 5.4. Figure 9. Percent depth doses for a 10×10 cm2 field of the ONCOR™ MV-CBCT, reported both in experimental dosimetry (blue line) and MC simulations (orange line). As for the Elekta Versa HD™ 6 MeV FFF beam, the highly available experimental data facilitated a deeper characterization of the beam. PDDs and profile measurements achieved great agreement for field sizes of 2×2 cm2 (Figure 10), 10×10 cm2 (Figure 11), and 20×20 cm2 (Figure 12). These results gave confidence for the following MC simulations in this work. The agreement obtained was essential since any small mismatch would result in some changes in the beam spectrum, therefore affecting the beam’s cross-section due to the presence of AuNP. MATERIALS AND METHODS 48 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy Figure 10. Elekta Versa HD™ beam characterization. The comparison between the experimental data (orange line) and the Monte Carlo simulation (blue line/dots) are shown for a percent depth dose and several profiles of a 2×2 cm2 field. MATERIALS AND METHODS José Antonio López Valverde – 49 Figure 11. Elekta Versa HD™ beam characterization. The comparison between the experimental data (orange line) and the Monte Carlo simulation (blue line/dots) are shown for a percent depth dose and several profiles of a 10×10 cm2 field. MATERIALS AND METHODS 50 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy Figure 12. Elekta Versa HD™ beam characterization. The comparison between the experimental data (orange line) and the Monte Carlo simulation (blue line/dots) are shown for a percent depth dose and several profiles of a 20×20 cm2 field. MATERIALS AND METHODS José Antonio López Valverde – 51 4.4.2. Gold nanoparticles virtualization, AuNP medium modeling, and simulation settings After modeling the geometry of the linear accelerators, the next step involved simulating particle transport from the linear accelerator to the patient/phantom using BEAMnrc [116, 117] user code, and calculating the dose in the patient with BEAMDOSE [120, 121], a modification of DOSXYZnrc [118, 119] developed by the Medical Physics group of Seville. This allows for understanding each beam’s contribution to a single voxel, a necessary step for treatment planning and optimization process. All these tools are implemented on the CARMEN platform [25–27]. Using this methodology, several clinical cases were simulated for this work, detailed later in section 4.4.4, where a certain amount of AuNP was presumed to be present in lesions previously defined as GTV/PTV on CT images or BTV on PET/CT images. Specifically, patient image data from clinical case plans was utilized by virtually adding nanoparticles to the BTV region, previously defined using specific SUVs present in the PET image, with the Foster segmentation algorithm [12]. Thus, the radiological concentration was homogeneously placed in the planning image in voxels labeled as part of the tumor, assuming a AuNP biodistribution efficiency similar to the PET capacity to identify the illness, which was considered a starting point for the methodology stated in section 4.1. Once the physical experimental procedures were finished, this approach should undoubtedly be studied under biological conditions using animal models, as discussed in section 4.6. Treatment planning in radiotherapy requires working with physical densities, since MC simulation considers the composition and cross-sections of the different materials distinguished in the CT conversion of the image. Consequently, material compositions were created in two different files: egs4phant and pegs4dat. The first one, egs4phant, is used to determine which voxel belongs to each material, based on its physical density. This file is generated from the clinical file using an in-house algorithm capable of converting from DICOM format, imported into the CARMEN platform [25–27]. It transforms HU into densities using CT calibration curves, as mentioned in section 4.3, and assigns materials to each physical density range. On the other hand, the second file, pegs4dat, is previously preprocessed for the physical MATERIALS AND METHODS 52 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy simulation. This file generates probabilistic data associated to the interactions of photons and electrons with different energies as they traverse the material mediums present in the egs4phant file. This preprocessing of cross-sections is a characteristic of EGSnrc that enhances efficiency by reducing the computational time required per simulation. Without it, simulations would take too long due to the large number of events needed to calculate macroscopic magnitudes, like dose values, with low statistical error. Two complementary tools were used for generating pegs4dat file: EGSnrcMP Graphical User Interface (GUI) version 0.9 [122], and pegs4.exe for parameters that cannot be configured inside that GUI. Five materials were generated: air, lung (with air), soft tissue, bone, and the new material that contains AuNP. This new AuNP egs4phant material was assigned only to the GTV/PTV or BTV region by interrupting the egs4phant-generating code and using the target structure as a mask to modify the material tag and the density of the voxels within that kernel. An illustration of AuNP placed in the context of the PTV of a clinical case of intact breast SABR is presented in Figure 13. Figure 13. 3D representation of a sample clinical case. Colors indicate the PTV with the new AuNP material virtualized (red), skin (gray), and chest wall (green). (A) Side and (B) front view. Thus, the relevant egs4phant file was generated by assigning each of the five materials and also, reducing the CT image resolution by half in the X and Y axes. This approach ensures that the statistical error remains acceptably low while significantly reducing simulation time. The material flags of the egs4phant file would point to the cross-sections of the material pre-calculated in the PEGS4 file. The compositions of the standard materials assigned are shown in Table 2. MATERIALS AND METHODS José Antonio López Valverde – 53 Table 2. Materials weight proportions used for computing the PEGS4 file, assuming it considered the significant digits rules: “air”, “lung (with air)”, “soft tissue” and “bone”, constructed respectively from “.element” default files “air_dry_nearsealevel”, “lung_inflated_icru_1986”, “tissue_soft_icrp” and “bone_compact_icru”. Element Air Density: 0.00120479 g/cm3 Lung (with air) Density: 0.26 g/cm3 Soft tissue Density: 1.00 g/cm3 Bone Density: 1.85 g/cm3 H - 0.103 0.10447 0.064 C 0.000124 0.105 0.23219 0.278 N 0.755267 0.031 0.02488 0.027 O 0.231781 0.749 0.63024 0.410 Na - 0.002 0.00113 - Mg - - 0.00013 0.002 P - 0.002 0.00133 0.070 S - 0.003 0.00199 0.002 Cl - 0.003 0.00134 - Ar 0.012827 - - - K - 0.002 0.00199 - Ca - - 0.00023 0.147 Fe - - 0.00005 - Zn - - 0.00003 - TOTAL 1 1 1 1 As for the new material, taking into account that the tumor can be radiologically considered as soft tissue, the AuNP material would be the result of adding the proper quantity of gold to the soft tissue material, with a similar physical density. At this point, three options for creating a material can be used: element, when there is only one type of atom implied in the material; compound, when there are different types of atoms, and the proportions of each atom are given in the relative number of atoms; and the final option is mixture, similar to the compound option but the proportions are given by mass. Following the reasoning above, mixture was the option applied for the new material containing AuNP. However, the real situation is not a homogeneous mixture of tissue and gold, but rather a heterogeneous dispersion of particles within the tissue, with discrete points – the nanoparticles – where the material is 100% gold, and points where the material is that of the soft tissue, with no gold at all. Nonetheless, such consideration in a voxelated simulation like the one proposed would need to be translated to nanometric voxel sizes. This approach is not computationally feasible for considering the whole macroscopic model. Additionally, several physical studies have shown that there are no differences in dose enhancement calculations between considering a heterogeneous MATERIALS AND METHODS 54 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy mixture and a homogeneous mixture, even when studying the microscopic physicochemical phase of the problem [123]. Thus, the material was modeled as a homogeneous mixture of gold in soft tissue, and the density of the voxels containing AuNP was increased to the maximum theoretical density – the original density plus the added gold nanoparticles concentration density – assuming that the space occupied by the AuNP and the weight of the water that previously occupied that space are negligible. The new AuNP material compositions for two different example concentrations are shown in Table 3. Those materials with the specific AuNP concentrations are some of the ones that were modeled, but only these two are shown for conciseness. Whenever different AuNP concentrations are mentioned in the text, it can be assumed that the same approach was followed. These two AuNP concentrations represent the ones most studied, considered to be the radiological and therapeutic concentrations, respectively (see sections 4.3, 4.4.3, and 4.4.4). Table 3. AuNP materials with different concentrations: weight proportions used for computing the PEGS4 file, based on the ‘soft tissue’ material. Element 2.23 mg Au/g of tissue Density: 1.00223 g/cm3 22.3 mg Au/g of tissue Density: 1.02230 g/cm3 H 0.10423903 0.10214227 C 0.23167222 0.22701216 N 0.02482452 0.02432518 O 0.62883257 0.61618369 Na 0.00112748 0.00110480 Mg 0.00012971 0.00012710 P 0.00132703 0.00130034 S 0.00198556 0.00194562 Cl 0.00133701 0.00131012 K 0.00198556 0.00194562 Ca 0.00022949 0.00022487 Fe 0.00004989 0.00004889 Zn 0.00002993 0.00002933 Au 0.00223000 0.02230000 TOTAL 1 1 Lastly, it is worth mentioning that with this preprocessing, besides generating interaction probabilities of photons and electrons with the material depending on its composition and density, other parameters related to the particle transport simulation are considered. In this study, the parameters involved were the following: radiative stopping powers (IAPRIM) activated, coherent/Rayleigh scattering (IRAYL) activated, additional stopping powers considerations (IUNRST) MATERIALS AND METHODS José Antonio López Valverde – 55 deactivated, lower charged particle cutoff energy (AE) 521 keV (considering the electron rest mass of 511 keV), lower photon cutoff energy (AP) 10 keV, upper charged particle cutoff energy (UE) 55 MeV, upper photon cutoff energy (UP) 55 MeV. More information on these parameters can be found in the relevant code documentation and manuals [119, 122]. Besides PEGS4 parameters, the dose calculation will depend on simulation parameters. These will have greater importance than usual in these simulations, described in detail in sections 4.4.3 and 4.4.4, because the differences caused by nanoparticles are expected to be tied to sensitive events, such as low energy interactions. The main parameters and corresponding values assigned to the BEAMDOSE user code [120, 121] are described in Table 4. Table 4. BEAMDOSE simulation parameters. BEAMDOSE parameter Value Global (lower) cutoff energy for charged particles (considering the electron rest mass of 511 keV) 0.521 MeV Global (lower) cutoff energy for photons 0.010 MeV Global maximum step-size restriction for electron transport 1010 cm Maximum fractional energy loss per step 1% Maximum first elastic scattering moment per step 0.1 Boundary crossing algorithm EXACT Skin depth for boundary crossing algorithm 0 Electron-step algorithm PRESTA-II Spin effects On Bremsstrahlung angular sampling KM Bremsstrahlung cross sections NIST Bound Compton scattering On Pair angular sampling KM Photoelectron angular sampling On Rayleigh scattering On Atomic relaxations On Electron impact ionization On The simulation parameters selected have been chosen to conduct precise calculations on the explicit transport of particles along the densities and materials included in the patient image, which is accompanied by a higher computational demand. Moreover, as mentioned, the influence of AuNP will manifest at the lowest energy levels, necessitating the selection of these parameters in a demanding scenario for dose calculation. Fortunately, our group has parallel computing resources, able to distribute complex tasks and therefore reduce elapsed time. MATERIALS AND METHODS 62 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy Table 6. Summary of the simulations performed. Clinical case(s) Energy beam AuNP concentration (mg Au/g of tissue) Head & Neck 6 MeV 0 6 MeV 2.23 6 MeV 40 MV-CBCT 0 MV-CBCT 2.23 MV-CBCT 22.3 Breast Cancer MV-CBCT 0 MV-CBCT 22.3 4.4.5. Dose enhancement experimental verification Although multiple MC simulations similar to the ones described have been previously verified by the research group where this thesis project was carried out, the inclusion of the new AuNP material prompts the need for an experimental dosimetric verification of the MC-based calculation of the dose enhancement. Using conventional ionization chambers is not considered appropriate because of several factors. Initially, the measurements are corrected based on factors that assume electronic equilibrium conditions. However, if the ionization chamber is placed inside a AuNP solution, corrections applied to the readings – based on constants and factors in water – might not be easily applicable to the AuNP medium, rendering them potentially inaccurate. For a first verification approach, we used the QuAArC phantom already mentioned in section 4.4.2. This phantom is made up of methacrylate, with all its components properly modeled and a known physical density. In this phantom, there are several spaces originally conceived for inserting a range of different ionization chamber models using properly designed inserts. Taking advantage of two of those inserts, we placed either water-equivalent DMEM or a AuNP solution at 22.3 mg Au/mL in DMEM inside the inserts’ cavities, and one small fragment of Gafchromic™ EBT3 radiochromic film inside each solution. The phantom was irradiated with the Siemens ONCOR™ MV-CBCT beam, using a treatment with twelve incidence angles and a symmetric 1×6 cm2 field for all the incidences. This treatment could be MC- simulated over QuAArC images where AuNP or water solutions were virtualized following the same methodology as in section 4.4.3, already described in detail in MATERIALS AND METHODS José Antonio López Valverde – 63 section 4.4.2. The relative DEF measured in both the experimental and simulated setups should match to validate the MC simulations. The irradiated dosimetric films were revealed using an Epson Expression 10000 XL (Seiko Epson Corp.) scanner, and the dose was measured using software developed by our research group collaborators in the MATLAB environment, using previously established protocols [6]. In parallel, a novel detector was designed in collaboration with the Radiophysics Laboratory of the University of Santiago de Compostela, led by Prof. Faustino Gómez Rodríguez. This detector was denominated “NanoLIC”, short for “nanoparticles liquid ionization chamber”, and it was inspired by previous designs by the Radiophysics Laboratory [125]. A phantom for hosting the chamber was also prototyped. In short, NanoLIC consists of a refillable detector that has isooctane as the active measurement medium. This would allow the measurement of doses in pure liquid isooctane and their comparison to the doses registered with AuNP solutions in isooctane, measuring dose enhancement directly in the AuNP media inside the chamber, and not next to it with an independent active medium. In other words, the AuNP solution would be the active medium. This chamber has been MC modeled by Dr. Elisa Jiménez by using egs_chamber user code [126], and the absorbed dose in the detection area was calculated both for pure isooctane, and AuNP at 22.3 mg Au/mL of solution in isooctane. The prototype of NanoLIC is shown along with its modeled computer-aided design (CAD) in Figure 15A,B. Based on those CAD files, a phantom for holstering nanoLIC and being able to stack it with solid water phantoms was designed and built. This phantom is depicted in Figure 15C,D. NanoLIC was tested with pure isooctane, synthesis grade (purity > 99.0%) [Sigma-Aldrich, Darmstadt, DEU, ref: 8.21627.1000] and spectroscopy grade (purity > 99.8%) [Supelco, Darmstadt, DEU, ref:1.04718.0500]. Dodecanethiol-functionalized AuNP, a hydrophobic formulation of gold nanoparticles presumably soluble in isooctane, have been acquired [Nanoprobes, MATERIALS AND METHODS 64 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy Yaphank, NY, USA, ref: 3014-80MG] but they have not been tested so far, as it will be discussed in section 5.4. Figure 15. NanoLIC and its phantom. (A) CAD design of NanoLIC; (B) The assembled prototype of the chamber; (C) CAD design of the phantom; (D) The manufactured NanoLIC-hosting phantom, along with NanoLIC placed inside. 4.5. Gold nanoparticles as radiosensitizers 4.5.1. Cell culture and irradiation Different cell lines were used depending on the radiotherapy modality. Cells were handled according to standard cell culture protocols and in accordance with the technical specifications for each cell line. The cells were maintained under controlled humidity, 37 °C temperature, and a 5% CO2 enriched atmosphere. For external beam radiotherapy, human breast cancer cells MDA-MB-231 [ATCC, Manassas, VA, ref: HTB-26] were the main ones used in this thesis project. They were grown in DMEM with high glucose and pyruvate [ThermoFisher, Paisley, GBR, ref: 21969035] supplemented with 4 mM L-glutamine [ThermoFisher, Paisley, GBR, MATERIALS AND METHODS José Antonio López Valverde – 65 ref: 25030081], 10% fetal bovine serum (FBS) [Sigma-Aldrich, Darmstadt, DEU, Ref: F7524-100ML], and penicillin (100 U/mL), streptomycin (100 µg/mL), and amphotericin B (0.25 µg/mL) [ThermoFisher, Grand Island, NY, USA, ref: 15240062]. For brachytherapy experiments, human cancerous uveal melanoma cells MP-46 [ATCC, Manassas, VA, ref: CRL-3298] were the main ones used. These cells were grown in RMPI 1640 medium [ThermoFisher, Paisley, GBR, ref: A1049101] supplemented with 20% FBS, and penicillin (100 U/mL), streptomycin (100 µg/mL), and amphotericin B (0.25 µg/mL). For experiments such as γH2AX immunofluorescence, the cells needed to be grown in coverslips. To promote cell adhesion to the coverslip surface, they were treated with poly-L-lysine [Sigma-Aldrich, Darmstadt, DEU, ref: P1524-25MG]. This treatment consisted of washing and flaming the coverslips using pure ethanol [Sigma-Aldrich, GBR, ref: 02851-1L] and sterile forceps for handling them. After the coverslips were washed and sterilized, and once they had cooled down, they were immersed for 16 hours in a 0.2 μm-filtered 0.1 mg/mL poly-L-lysine solution prepared in Trizma™ buffer at pH 8.5 using Trizma™ base [Sigma-Aldrich, St. Louis, MO, USA, ref: T1503-25G] and Trizma™ hydrochloride [Sigma-Aldrich, St. Louis, MO, USA, ref: T3253-100G]. Finally, the coverslips were rinsed with 0.2 μm-filtered milli- Q water, then with a 0.2 μm-filtered solution of 0.9% w/v NaCl solution, and then again with 0.2 μm-filtered distilled water. For these γH2AX experiments, performed both for external beam radiotherapy and brachytherapy, the cells used were in their fourth passage. They were grown in 12 mm circular coverslips treated with poly-L-lysine and placed on the bottom of 24-well cell culture plates. The cells were grown for 24 hours to ensure that they had time to attach and recover from the subculture procedure. For external beam radiotherapy experiments, the four central wells in five plates were used to grow the cells, totaling 20 wells. In each plate, two wells contained complete DMEM, while the other two wells contained complete DMEM and 15 nm AuroVist AuNP at ~22.3 mg Au/mL of media, our proposed AuNP concentration for external beam radiotherapy obtained in the previous stages. For brachytherapy experiments, the eight central wells in four plates were used to grow the cells, totaling 32 wells. In MATERIALS AND METHODS 66 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy each plate, there were 4 different conditions: DMEM, and AuNP solutions in DMEM at concentrations of 0.05, 0.2, and 1 mg Au/mL of solution. After the nanoparticles were added, an incubation of 12 hours took place for external beam radiotherapy experiments. For brachytherapy experiments, this incubation period was reduced to 30 minutes, given that the nanoparticles would remain in the culture during the complete irradiation procedure (~24 hours). For micronuclei assay, only conducted for external beam radiotherapy, the cells used for the experiments were in their ninth passage. They were grown in 24-well cell culture plates and seeded at a density of 100 000 cells/well. Again, the four central wells in five plates were used to grow the cells, totaling 20 wells. The cells were grown for 24 hours, assuring that they had proper time to attach and recover from the subculture procedure. After that period, the media were. In each plate, two wells were incubated 2 more hours with either 500 µl of complete DMEM or with complete DMEM containing 15 nm AuroVist AuNP at ~22.3 mg Au/mL of media. After the incubation step, the cells were irradiated. For external beam radiotherapy experiments, the plates were placed on the treatment table and irradiated using Siemens Oncor MV-CBCT IGRT beam. Each cell plate constituted a different irradiation dose condition for the study. The plates were irradiated from the bottom to avoid air gaps in the incidence path, and they were placed on top of 1 cm of solid water phantom, at the maximum of the PDD for that beam spectrum. Solid water was also placed on top of the cell plate lids, to reduce the impact of a possible build-down effect. An image of the irradiation setup for external beam radiotherapy is shown in Figure 16. Cells were irradiated with 0.25 Gy, 0.5 Gy, 1 Gy, and 2 Gy. A plate remained unirradiated, as a control plate. 30 minutes after irradiation, the media in the wells, either with or without nanoparticles, was discarded, and the cells underwent three PBS [Sigma-Aldrich, St. Louis, MO, USA, ref: P4417-50TAB] washes. This is the starting point for both the γH2AX immunofluorescence protocol and the micronuclei protocol. MATERIALS AND METHODS José Antonio López Valverde – 67 Figure 16. Cells irradiation setup for external beam radiotherapy with the MV-CBCT beam. For brachytherapy experiments, the cell plates were placed on an incubator at HUVM where the cells would be irradiated for 23 hours there to maintain optimal temperature, CO2, and humidity conditions for the cells during the whole irradiation procedure. Similarly to what was previously reported by Ngwa et al. [127], 125I seeds were placed in a custom-made phantom that was designed by the research group member Héctor Miras del Río, assisted by MC simulations, to ensure that the dose to the 8 central wells of a 24 well plate was as homogeneous as possible (<5% variation over all the surface of interest). The irradiation setup is depicted in Figure 17. Different dose rate 125I seeds were used to irradiate the cells for 23 hours, obtaining doses of 0.76, 1.73, and 3.10 Gy. A control plate remained unirradiated. One hour after the irradiation, the cell media with or without nanoparticles were discarded, and the cells underwent three PBS washes. Fresh medium was added, and the cells were left to incubate for 24 more hours to allow for the DNA damage to be repaired and observe unrepairable damage. This method also normalizes the differences between the DNA damage occurring early in the irradiation process and damage occurring nearly 23 hours later, at the end of the irradiation. MATERIALS AND METHODS 68 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy Figure 17. Brachytherapy irradiation phantom for positioning the 125I seeds and placing the cell plate. (A) Top view of the mold with the gaps where the seeds are placed. (B) Picture with the lid of a cell culture plate for illustrating the placement of the cells for irradiation in the incubator. Pictures courtesy of Héctor Miras. For both types of experiments, external beam radiotherapy with MV-CBCT and brachytherapy, the distribution of the different cell conditions along the radiation profile of the beam was considered. This way, for MV-CBCT, the two samples of each condition (AuNP and control) were placed diagonally (see detail in Figure 16), and for brachytherapy, the samples of each condition (the three AuNP concentrations and control) were placed such that each condition had one sample in one of the four innermost wells and the other sample in the four outermost wells. 4.5.2. Immunofluorescence assay for evaluating DNA damage – γH2AX foci 4.5.2.1. Immunofluorescence protocol For observing early DNA damage, the cells were fixed 30 minutes after irradiation. For late (unrepairable) DNA damage, the fixation step was conducted 24 hours after irradiation. Fixation was performed using a 4% paraformaldehyde [Sigma-Aldrich, Darmstadt, DEU, ref: P6148-500G] solution in PBS for 10 min. Then, the cells were washed three times for 5 min in agitation and permeabilized using a 0.1% Triton-X 100 solution [Fisher Bioreagents, Madrid, Spain, ref: BP151-500] for 10 min. Next, a blocking solution consisting of Triton X-100 0.05% and BSA 3% [Sigma-Aldrich, Darmstadt, Germany, ref: A9647-50G] was applied for 30 min. After this preparation, a primary antibody targeting the phosphorylated Ser-139 of H2AX [Novus, Madrid, ESP, ref: NB100-74435] was added at a concentration of 2 µg/mL in a PBS solution along with Triton X-100 0.05% and BSA 3%. Following 2 h of incubation at r.t., the coverslips were washed three times for 5 min in agitation using a PBS solution with Triton X-100 0.05% + BSA 3%. MATERIALS AND METHODS José Antonio López Valverde – 69 Subsequently, incubation with the secondary antibody Alexa Fluor 488 [Invitrogen, Eugene, OR, USA, ref: A-21121] took place, at a 1:1000 concentration in a PBS solution with Triton X-100 0.05% and BSA 3%. After one hour of incubation at r.t., the coverslips were washed three times with PBS for five minutes in agitation and mounted with Fluoromount + DAPI [Invitrogen, Carlsbad, CA, USA, ref: 00-4959- 52]. 4.5.2.2. Imaging and quantification Each preparation was scanned with the Leica Stellaris 8 confocal microscope using the stimulated emission depletion (STED) super-resolution technique. All images were taken with the HC PL APO CS2 100×/1.40 OIL objective. The microscope used is shown in Figure 18. Figure 18. Leica Stellaris 8 confocal microscope used for the imaging of the γH2AX foci with STED technique. Image courtesy of the University of Seville research central services. A sequential scan was made, saving fluorescence signal information in two channels. The blue channel was assigned to the DAPI signal corresponding to the cell nucleus, with the pinhole set at 1 AU, excitation at 405 nm with intensity ~10%, and a detector HyD S 1 (430–499 nm) with the gain set to 24. The green channel MATERIALS AND METHODS 70 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy recorded the information of the γH2AX foci, with the pinhole set at 1 AU, excitation at 499 nm with intensity ~6.7%, and a detector HyD S 3 (504–581 nm) with the gain set to 16. For the foci channel, STED was used, with a 592 nm depletion laser with intensity set to 20%. A wide tile scan composed of at least 16 single image fields was taken for each sample, obtaining 2 tile scan images per irradiation condition, both for control and nanoparticle-treated cells. Each experimental condition consisted of at least 200 cellular objects and up to 360 objects. The pixel resolution of the single image fields was optimized according to Nyquist’s law. The resolution for each image constituting the tile scan was 1880×1880 px, and the pixel depth was set to 8 bits for each channel. The signal detected in the γH2AX channel was averaged from 3 different scans, to reduce signal noise. 4.5.2.3. Image processing For this study, we developed a robust automatic quantification software, which we called SoFoCo, an adaptation of the FoCo algorithm by Lapytsko et al. [128]. Unlike FoCo, our SoFoCo software [v1.0, Seville, Spain] was coded entirely in MATLAB. Also, the DNA damage quantification levels are not measured in terms of the absolute number of foci per cell nucleus, which is an operation highly dependent on the nucleus size, on the height at which the Z-slice was taken in the confocal microscopy, and on cell labeling algorithm used to separate cell groups into individual cells. Rather, we quantified the foci density per object, whether they are individual cells or small groups not separated by thresholding. This way, some steps are modified or suppressed, such as the usage of watershed algorithm for labeling individual cells, or the need to eliminate objects at the borders of the image. An illustrative workflow of the followed process is available in Figure 19. MATERIALS AND METHODS José Antonio López Valverde – 71 Figure 19. Workflow of the SoFoCo algorithm used for foci density quantification [129]. In short, this image processing workflow involves separating the foci and nuclei channels. In the nuclei channel, a threshold is applied to binarize the image, in our case with the Otsu method [130]. Then, an adaptive median filter is applied to denoise the image and holes in the objects are filled. Following this, a morphological opening by reconstruction deletes small objects that are not considered nuclei, using a minimal nuclei radius of 25 pixels for our image dataset. After that, a morphological closing filter is applied by dilating, filling holes, and eroding. The dilations and erosions were performed 4 times. The objects in that last mask are used as the individual areas in which the foci density will be counted. To count the MATERIALS AND METHODS 78 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy Figure 21. Pictures, illustrations, and schemes of the simulation setup. (A) Picture of the placement of the cells for irradiation and (B) its representation in the modeling. (C) Detail of the cell plate placement and (D) cell plate model detail. (E) Scheme of the geometries at the second stage of the irradiation. MATERIALS AND METHODS José Antonio López Valverde – 79 4.6. Animal models experimentation – ethics committee proposal Parameters such as long-term toxicity or AuNP biodistribution are set to be evaluated using in vivo experiments. Those experiments will consist of injecting several AuNP concentrations in adult rats and observing their response. Once toxicity assays have shown good results, the next step will consist of the AuNP-CT vs PET-CT comparison, using in vivo organisms with cancer, such as rats with induced mammary carcinoma. If the results are good, we would evaluate the same approach in bigger animal models, canine cancer patients, that could be treated and this way it would be possible to assess biological response to the use of AuNP as theragnostic agents in RT. Despite the whole project is too ambitious to be carried out in the calendar established for this thesis project, the ideas exposed in the previous paragraphs were drafted in the form of application required by the ethical committee of University of Seville and the corresponding regional institution to be accepted for an experimental project with animals in accordance with the current legislation (see supplementary material section 8.7). MATERIALS AND METHODS 80 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy RESULTS AND DISCUSSION José Antonio López Valverde – 81 5. Results and discussion 5.1. Nanoparticle synthesis The synthesis procedure was successfully carried out, yielding reproducible results with minor variations in the absolute concentration. An example of the absolute and the normalized spectra of two different independent syntheses can be seen in Figure 22A and Figure 22B, respectively. These syntheses are independent not only batch-wise and timewise, but also in terms of the material used for the synthesis, as a different set of glassware with different sizes was used for escalating the synthesis. Even different heating plates were utilized. The main parameters of the spectra measured can be found in Table 7. The fitting methodology described for calculating the parameters of the spectra yielded R2 values higher than 0.999. Table 7. UV-Vis spectra characterization based on the fitted UV-Vis spectra. Spectra Plasmon peak (nm) OD at peak Dispersion Disp95 Disp90 Disp85 Synthesis A 520.12 1.028 0.0445 0.0644 0.0973 Synthesis A + 15 days (4 °C) 522.07 1.037 0.0429 0.0624 0.0949 Synthesis A + 5 days (r.t.) 521.17 0.980 0.0421 0.0614 0.0936 Synthesis B 520.27 0.967 0.0436 0.0632 0.0958 Synthesis B + 15 days (4 °C) 521.32 0.982 0.0429 0.0628 0.0959 Synthesis B + 21 days (4 °C) 521.32 0.995 0.0423 0.0616 0.0945 As can be seen in Figure 22 and in Table 7, no significant change between different syntheses was found, as long as the reagents’ molar ratio was respected, in line with the conclusions from the synthesis protocol [108]. The DLS measurements found three AuNP size distributions: hydrodynamic diameters around 28 nm (~76%), 160 nm (~17%), and 1 nm (~7%). While the first and third distributions are expected according to the synthesis protocol considering the chosen sodium citrate to chloroauric acid molar ratio of approximately 2.8 [108], the second one was not expected. However, TEM measurements discarded the presence of such particles. This second distribution was most likely reported due to an experimental measurement artifact or could hint at the presence of temporal dispersible associations between nanoparticles after vortexing and sonication. RESULTS AND DISCUSSION 82 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy Figure 22. UV-Vis spectra of two solutions of AuNP from independent syntheses, with different representations: (A) complete absolute UV-Vis spectra, and (B) a detail range (450-600 nm) of the normalized UV-Vis spectra containing the surface resonance peak. RESULTS AND DISCUSSION José Antonio López Valverde – 83 The average core diameter of the nanoparticles measured was around 23 nm in high accordance with the reference protocol. This is in line with the DLS results, as DLS reports hydrodynamic diameters, accounting for the organic shell of the AuNP, whereas the TEM contrast only allows for the detection of the core of the AuNP thanks to gold’s higher electronic density, compared to citrate. Sample TEM images are shown in Figure 23A-C, where a relatively stable AuNP size was found, in contrast with the findings for commercial AuNP, depicted in Figure 23D-F. The size distribution characterization of the sample synthesis lot given has an average AuNP core diameter of 23.18 nm and a standard deviation of 1.87 nm. The size distribution of this lot is presented in Figure 24. The data distribution was not statistically different from a normal distribution, according to the Shapiro-Wilk test. Figure 23. TEM micrographs of AuNP at magnifications ranging from 50000× to 200000×. The upper row (A-C) represents sample images of the nanoparticles synthesized by us. The lower row (D-F) represents sample images of commercial AuNP AuroVist™-15 nm. It is worth noting that in Figure 23A-C, there are no visible nanoparticles with diameters within the nanometer range, contrary to the DLS measurements. This is probably related to two causes: on the one hand, most of those particles are discarded after centrifugation, as they tend to remain in the supernatant fraction; on the other hand, a hydrodynamic diameter of 1 nm means sub-nanometric core RESULTS AND DISCUSSION 84 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy diameters, which could fall close to the theoretical resolution of the TEM microscope, and most importantly, those AuNP might not produce enough contrast to be easily distinguishable in comparison to the formvar/carbon matrix of the grid where the sample is placed for TEM imaging. Also, nanoparticles clustering in TEM microscopy images such as the ones shown in Figure 23 do not necessarily correspond to aggregation in the original AuNP solution, given that the TEM sample preparation process forces conditions such as drying the particles, potentially priming aggregation of previously-unaggregated AuNP. Figure 24. AuNP size distribution histogram of a sample synthesis lot (blue). The average core diameter is 23.18 nm, with a standard deviation of 1.87 nm (n=306). The data distribution was not statistically different from a normal distribution. A normal distribution is represented with the average and standard deviation AuNP core diameter (red). The AuNP concentration estimation result of a given synthesis, according to the weight of the solid residue after drying the particles solution by complete evaporation was around 0.068 mg Au/mL of solution. However, this theoretical maximum concentration of the synthesis solution assumes that all the weight of the solid residue is gold, not accounting other molecules such as citrate, or residual water. In fact, this concentration would consider a larger amount of gold than the quantity initially added to the synthesis. Hence, it is no surprise that the real gold concentration, as reported by ICP-OES, is around 0.045 mg Au/mL of solution. This would give a reaction efficiency over 90%. RESULTS AND DISCUSSION José Antonio López Valverde – 85 The theoretical re-concentration factor, assuming no AuNP are lost when discarding the supernatant phase after centrifugation, is around 90×. The centrifugation was carried out in 50 mL centrifuge tubes, and the concentrated AuNP fractions can be combined to get a larger volume, but they are not stable enough to undergo further centrifugation when combined. Further centrifugation results in aggregates and precipitates of the particles. Centrifugations with 15 mL centrifuge tubes did not provide better results, despite being geometrically more favorable as they present a smaller angle change on the conic section of the tubes, compared to the 50 mL tubes. Hence, the maximal achievable concentration by centrifugation under these conditions is around 4 mg Au/mL of solution. To proceed with further centrifugation, a stabilization of the AuNP should be investigated to prevent aggregation. Concentrating the AuNP through complete rotary evaporation was proven not feasible, as the resulting solid residue was not dispersible. Partial rotary evaporation could be researched in the future to continue concentrating the AuNP beyond the aforementioned concentration threshold, but further stabilization of the nanoparticles might be necessary for this method as well. Another option to test in the future would be concentrating the nanoparticles based on centrifugation but, instead of using high speeds, using centrifugal filters and lower speeds to reduce physical stress applied to the nanoparticles. This approach was later found to be used by some commercial AuNP products, such as AuroVist, which was further concentrated in a study by Hainfeld et al. using a 100-kDa centrifugal filter [85]. While concentrations on the order of the few mg Au/mL are suitable for many purposes, some other applications might need even higher concentrations. Commercial nanoparticles, such as the ones already depicted in Figure 23D-F, are supplied at concentrations up to 200 mg Au/mL of solution. The large reconcentration factor that these AuNP undergo from the original synthesis concentration could be the reason why their size variability is higher than the one obtained for the particles newly synthesized for this thesis project. In addition to a reproducible and scalable synthesis with an appropriate size distribution, the resulting AuNP were proven to be stable over time, both in the medium term when they were stored at 4 °C, and in the short term when at room temperature. Figure 25 shows a detail of the surface plasmon resonance peak for RESULTS AND DISCUSSION 86 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy the same example synthesis A & B already shown in Figure 22, and different time lapses for AuNP stored at 4 °C (15 and 21 days), and AuNP stored at room temperature (5 days). The parameters of each spectrum are also reported in Table 7. The small differences found fall within the range of the experimental measurement uncertainty. Thus, the lack of significant differences between the different spectra hints that no substantial differences are expected in the AuNP solutions in terms of particle size distribution. Figure 25. Detail range (450-600 nm) of the normalized UV-Vis spectra containing the surface resonance peak of the synthesis of two synthesis: synthesis A (blue line) and its follow up measurements after 15 days at 4 °C (yellow line) and 5 days at room temperature (pink line); synthesis B (orange line) and its follow up measurements after 15 days at 4 °C (purple line) and 21 days at 4 °C (green line). The nanoparticles can be considered stable enough to be used in the proposed methodology. They have demonstrated stability for three weeks when stored in the fridge. Moreover, they are robust enough to withstand temperature fluctuations during transportation, as evidenced by their stability at room temperature for a duration of 5 days. These time lapses were studied because they were long enough RESULTS AND DISCUSSION José Antonio López Valverde – 87 for the completion of the whole methodology described. However, a future stability study followed through a longer time frame might generate key insights on the complete interval where the AuNP are found to be stable at different conditions. The synthesis and characterization of AuNP was overall a success, even more considering the circumstances when it was carried out. This part of the experimentation was heavily affected by the global COVID-19 pandemic events, forcing pauses in the experimentation multiple times, some of them for long durations. However, even though the results were positive in general and the AuNP could be used for several physics experiments in the early stages of the thesis project such as testing the contrast power of gold in CT images, revisiting this research line in the near future to carry further experimentation would be an interesting path to continue increasing the concentration of the AuNP and, more importantly, testing the toxicity and biocompatibility profiles of these newly synthesized AuNP. Due to the timeline of the thesis project being heavily disrupted, and due to complications linked to new potential periods of quarantine imposed by the COVID- 19 pandemic, it was decided to use commercial gold nanoparticles for applications where the thesis project needed higher AuNP concentrations, and for biological experiments where AuNP biologically tested and found non-toxic were needed. In any case, this initial phase devoted to our in-house synthesis of nanoparticles can be considered crucial for this work. It not only allowed us to critically evaluate the commercially acquired product from Nanoprobes but also facilitated our transition into an advanced customer. We even engaged in correspondence with the technical services of the spin-off established by Hainfeld in New York. This communication involved the exchange of characterization data for their products, contributing to a more robust quality control of the Aurovist™ nanoparticles utilized in the remaining experiments. 5.2. Gold nanoparticles as contrast agents The CT and MV-CBCT images of the electron density phantom with AuNP placed in the center insert slot were acquired. The CT data obtained suggest that energies such as 100 or 120 keV are preferable to 80 or 140 keV. This may be due to 80 keV being slightly lower than the gold K-edge, whereas 100 and 120 keV beams particles RESULTS AND DISCUSSION 94 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy Figure 30. Comparative sample MV-CBCT axial slices of NAOMI phantom containing (A-C) waterequivalent DMEM or (D-F) gold nanoparticles at ~21 mg Au/mL of solution. Images are shown (A, D) without adjusting window level – i.e., complete image range – and (B, E) after adjusting the window level. A detailed view of them rather than the complete field of view is also shown in images C, F. Nonetheless, it is important to note that even the contrast levels observed with MV-CBCT permitted an automatic three-dimensional segmentation of the region of interest, obtaining structures with a number of voxels (2.861 mm3 per voxel) equivalent to a volume almost identical to the known original AuNP volume contained in the Eppendorf. An example of the volume calculation based on region growing segmentation algorithm for the image after applying a window level is shown in Figure 31. For that figure, different gray thresholds were used for the region growing algorithm, obtaining a dataset of voxelized volumes that stabilizes where the threshold are optimal – i.e., they are not too low and part of the target volume is not included in the structure, nor they are too high and include undesired volumes. After fitting the dataset to a cubic polynomial curve, the inflection point was calculated where the derivative of the original curve is 0, and a volume of 2011 mm3 was found. This volume is almost the real volume of 2 cm3 included in the vial. The differences between the real volume and the volume of interest calculated in the image using the region growing algorithm might be due to experimental error such as pipetting error, due to signal noise, or due to the resolution of the image weighing the high HU of AuNP along with the low HU of the Eppendorf plastic in the bordering voxels of the structure. RESULTS AND DISCUSSION José Antonio López Valverde – 95 Figure 31. Volumes of the automatically segmented AuNP structures with different region growing threshold values. The fitting goodness R2 parameter is presented along the fitting curve equation. Therefore, this beam could be proposed for IGRT and real-time ART, serving both as a therapeutic beam and providing live functional imaging of the tumor region contrasted with AuNP in the treatment room. Thus, this adaptive radiotherapy approach could be used as an efficient personalized radiotherapy. More on the performance of an experimental therapeutic beam with this spectrum is discussed in the sections 5.3 and 5.3.2. In general, even though AuNP have a high value as a contrast agent at the higher concentrations evaluated, this contrast capacity remains less powerful than previous results in the scientific literature. The main reason for this disagreement is that clinical conditions were applied in this work, whereas in most of the published literature, AuNP contrast is measured using nano-CT, or acquisitions lack the positioning of AuNP in a clinical context, i.e., AuNP not surrounded by the tissueequivalent medium. Therefore, the contrast power of AuNP is ensured, but the concentration required is higher than the one proposed in previous works. This new AuNP concentration could compromise its pharmacokinetic and toxicological possibilities unlike other lower concentrations used before in rodents within nano- CT, but the size differences between these animals and humans allows to be optimistic at this point. Previous studies on the toxicological effects of these AuNP have not shown toxicity issues at these concentrations. Hainfeld et al. state that they have not found any toxicological effect and, so far, the only limitation they have found on upscaling the concentration of the injections is due to the maximum intravenous injectable volume that mice can take [107]. On some studies already RESULTS AND DISCUSSION 96 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy mentioned in the Introduction section, they report using injections of 270 mg Au/mL [86], and even 400 mg/mL for injecting 4 g Au/kg of mouse [85], without observing any toxicological adverse effect. We expect that the proposed concentration values can be achieved locally in tumor cells considering the low toxicity of the product, with an LD50 > 5 g Au/kg, according to the nanoparticles’ data sheet [107]. Moreover, the numerous possible AuNP functionalizations, augmenting their targeting efficiency, encourage further efforts to transfer these results to the clinic. Given AuNP’s long blood circulation times with a stated half-life of between 15 and 24 h [107], sufficient time is allowed for treatment planning after image acquisition and even for acquiring follow-up images after treatment. Nebuloni et al. stated that an anatomical representation of blood vessels could be possible at 2 and 3 days after injection in mice [139]. These findings generate expectations of using MV-CBCT for follow-up imaging without the need for additional injections. Additionally, if AuNP were used solely for diagnostic purposes, rather than for real-time IGRT/ART, exploring their performance in kilovoltage-based CBCT would be worthwhile. Such systems with lower energies are expected to yield better results than MV-CBCT due to the greater component of photoelectric absorption. In any case, it would be useful to compare the performance of these functional images with PET imaging, today’s gold-standard. Future work is already scheduled in our project’s roadmap, using animal models, as discussed in section 4.6. 5.3. Gold nanoparticles as dose enhancers 5.3.1. Dose enhancement dependence on key parameters As it was discussed in section 4.4.3, these key parameters are interdependent. In this sense, although the results aim to evaluate the influence of each parameter by itself, a discussion was followed considering a global framework. The results of the dose enhancement factor for the volumes defined in the QuAArC phantom CT image (see section 4.4.3) for simulations based on the Siemens ONCOR™ MV-CBCT beam and different nominal energies of the incident beam adapted from Elekta Versa HD™ linear accelerator are shown in Table 9 and summarized in Figure 32. RESULTS AND DISCUSSION José Antonio López Valverde – 97 Table 9. Results of the simulation of dose enhancement over the QuAArC phantom with virtualized AuNP at 10 cm depth, with Siemens ONCOR™ MV-CBCT beam, and different nominal energies of the incident beam on an Elekta Versa HD™ linear accelerator. “Adjust.” stands for a field size MLC-shaped to the size of the target, approximately a 2×2 cm2 field. MC error refers to the maximum absolute statistical error of the Monte Carlo simulation. Energy (MeV) PDD max depth (mm) K-factor (1/MC dose at PDD max depth) Field size (cm2) DEF ± MC error Target Target center Outer ring 1 0.25 1.06E+16 10×10 1.22 ± 0.05 1.23 ± 0.05 1.00 ± 0.04 Adjust. 1.08 ± 0.04 1.09 ± 0.04 1.00 ± 0.04 1.5 0.35 2.94E+15 10×10 1.15 ± 0.04 1.15 ± 0.04 1.00 ± 0.04 Adjust. 1.05 ± 0.04 1.05 ± 0.04 1.00 ± 0.04 2 0.55 1.20E+15 10×10 1.11 ± 0.04 1.12 ± 0.04 1.00 ± 0.04 Adjust. 1.03 ± 0.03 1.04 ± 0.03 1.00 ± 0.04 2.5 0.55 6.09E+14 10×10 1.08 ± 0.04 1.09 ± 0.04 1.00 ± 0.04 Adjust. 1.02 ± 0.03 1.02 ± 0.03 1.00 ± 0.03 3 0.85 3.57E+14 10×10 1.07 ± 0.04 1.08 ± 0.04 1.00 ± 0.04 Adjust. 1.02 ± 0.03 1.02 ± 0.03 1.00 ± 0.03 3.5 0.85 2.43E+14 10×10 1.06 ± 0.03 1.07 ± 0.03 1.00 ± 0.03 Adjust. 1.02 ± 0.03 1.02 ± 0.03 1.00 ± 0.03 4 0.85 1.85E+14 10×10 1.05 ± 0.03 1.06 ± 0.03 1.00 ± 0.03 Adjust. 1.01 ± 0.03 1.01 ± 0.03 1.00 ± 0.03 MVCBCT 1.05 1.70E+14 10×10 1.20 ± 0.03 1.20 ± 0.03 1.00 ± 0.03 Adjust. 1.07 ± 0.02 1.07 ± 0.02 1.00 ± 0.02 Figure 32. Representation of the dose enhancement factors obtained for 10×10 cm2 (blue bars) and MLC-shaped (~2×2 cm2, orange bars) fields used to irradiate the QuAArC phantom with virtualized AuNP, with a Siemens ONCOR™ MV-CBCT beam and different nominal energies of the incident beam on an Elekta Versa HD™ linear accelerator. The error bars represent the maximum absolute statistical error of the Monte Carlo simulation. RESULTS AND DISCUSSION 98 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy The dose enhancement was shown to be an exclusively local effect, only affecting the voxels where AuNP were virtualized. This is captured in Table 9 on the rightmost column, where the DEF on the “ring” structure around the virtualized volume of AuNP was consistently reported as 1.00 – i.e., no change was found in the dose calculated in that structure when the AuNP were present in the phantom compared to when they were not. Also, there was no substantial difference between the DEF in the “target” structure and the “target center” structure, containing only the central voxels of the target structure. Therefore, there were no evident neighboring effects for this small volume. However, a slight non-significant tendency to observe a small increase in the DEF for the “target center” compared to the whole “target” structure was found, mainly for the Versa-based simulations. Studying this possible effect in different conditions, such as larger target volumes, could be an interesting resource for future research. Besides the relative effects of the DEF between structures for each energy, the impact of using different energies was observed. The DEF for the virtualized AuNP structure – “target” structure – ranged from marginal values up to increases of 1.22 and 1.08 in the dose on the target for the 10×10 cm2 and the MLC-shaped (~2×2 cm2) fields, respectively. An inverse correlation between DEF and the mean nominal energy of the incident beam for Elekta Versa HD™ was observed. This was in line with the expected results, given that a softer energy spectrum is associated with a higher photoelectric component of the beam, and therefore, a higher dose enhancement. The MV-CBCT represents a special case; despite its mean nominal energy being 4.2 MeV, it achieved DEF results of 1.20 and 1.07 for the 10×10 cm2 and the MLC-shaped (~2×2 cm2) fields, respectively. These results are closer to the 1 MeV spectrum than to any other energies simulated on the Elekta Versa HD™. This was likely caused by the presence of a carbon-based target in the Siemens ONCOR™ MV-CBCT beam, which yields an energy spectrum with a mean energy of around 1.4 MeV at the isocenter level. However, as it has been mentioned, the DEF of the MVCBCT is more similar to the DEF obtained with the energy spectrum of 1 MeV on the Versa HD™ than with the 1.5 MeV energy spectrum. This could be due to the beam’s RESULTS AND DISCUSSION José Antonio López Valverde – 99 softer spectra, attributed to the higher participation of low energies in the MV-CBCT, which are expected to be the main contributors to the dose enhancement factor. It is worth noting that for each energy, both for the Versa HD™ set of energies and the MV-CBCT beam, the DEF achieved with the 10×10 cm2 field is higher than the one obtained with the MLC-shaped (~2×2 cm2) field. This effect, already described by Detappe et al. for gadolinium-based nanoparticles [36], is probably related to an increased proportion of low energy secondary particles from bigger fields. This phenomena is directly related to the field size, and not to the number of incidences used, as no difference was observed, as expected, when varying the number of angular incidences for the MV-CBCT beam with the MLC-shaped field size (~2×2 cm2), as shown in Table 10. Table 10. Results of the simulation of dose enhancement over the QuAArC phantom with virtualized AuNP at 10 cm depth and multiple incidences, using the MV-CBCT beam and an MLC-shaped field size (~2×2 cm2). MC error refers to the maximum absolute statistical error of the Monte Carlo simulation. Number of incidences DEF ± MC error Target Target center Outer ring 4 1.07 ± 0.02 1.07 ± 0.02 1.00 ± 0.02 8 1.07 ± 0.02 1.07 ± 0.02 1.00 ± 0.02 16 1.07 ± 0.01 1.07 ± 0.01 1.00 ± 0.02 The simulation of different field sizes confirmed this trend: a smaller field size correlated with a smaller DEF, as shown both in Table 11 and Figure 33. Table 11. Results of the simulation of dose enhancement over the QuAArC phantom with virtualized AuNP at 10 cm depth and different field sizes for irradiating with a 1 MeV mean energy spectrum over an Elekta Versa HD™ linear accelerator. MC error refers to the maximum absolute statistical error of the Monte Carlo simulation. Field size (cm2) DEF ± MC error Target Target center Outer ring 10×10 1.22 ± 0.05 1.23 ± 0.05 1.00 ± 0.04 5×5 1.14 ± 0.05 1.15 ± 0.04 1.00 ± 0.04 3×3 1.09 ± 0.04 1.11 ± 0.04 1.01 ± 0.04 MLC-shaped (~2×2) 1.08 ± 0.04 1.09 ± 0.04 1.00 ± 0.04 RESULTS AND DISCUSSION 100 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy Figure 33. Representation of the dose enhancement factors obtained for different field sizes used to irradiate the QuAArC phantom with virtualized AuNP at 10 cm depth, with a spectrum of 1 MeV mean energy over an Elekta Versa HD™ linear accelerator. The error bars represent the maximum absolute statistical error of the Monte Carlo simulation. The simulations using the “column configuration” of the phantom (see Figure 14C), featuring only one beam incidence at 0 degrees, as detailed in section 4.4.3, confirmed that the observed effect was due to scattered radiation within the phantom. This contributed to a higher DEF when lower-energy particles, originating outside the nanoparticles region, reached the AuNP “target” structure volume. These results are shown in Table 12. Table 12. Results of the simulation of dose enhancement over the QuAArC phantom with virtualized AuNP at 10 cm depth and different QuAArC phantom configurations, irradiating with a 1 MeV mean energy spectrum over an Elekta Versa HD™ linear accelerator with a 10×10 cm2 field size. MC error refers to the maximum absolute statistical error of the Monte Carlo simulation. Phantom DEF ± MC error Target Target center Outer ring Whole phantom 1.23 ± 0.06 1.24 ± 0.06 1.00 ± 0.05 “Column configuration” 1.07 ± 0.05 1.07 ± 0.05 1.00 ± 0.05 Although the errors in the table above are larger due to the use of only one incidence, the DEF values were slightly greater for the whole phantom configuration compared to those reported for the original simulation of the 1 MeV mean nominal energy in the Elekta Versa HD™ with the 10×10 cm2 field size, as previously reported in Table 9. Given that the sole difference was the substitution of air by vacuum, as explained in section 4.4.3, this suggests a possible filtering effect of air, reducing the RESULTS AND DISCUSSION José Antonio López Valverde – 101 number of low-energy particles that reach the AuNP “target” structure. However, the effect of air on the beam spectrum is limited compared to that of tissue or other materials. Consequently, a more significant effect was observed when changing the depth of the “target” structure, using the “mini QuAArC configuration” previously illustrated in Figure 14C. Hence, an increase in DEF for the shallower AuNP “target” structure was observed, as shown in Table 13 and its comparison with Table 9, summarized in Figure 34. The difference between target depths was bigger for lower energies, as the increase in DEF was likely promoted by the low energy component of the beam. This is the most plausible reason why the MV-CBCT beam outperformed all the explored possibilities of the Elekta Versa HD™ theoretical beams, obtaining the maximum DEF of 1.14 with the MLC-shaped field size. For 10 cm depth, the 4.2 MeV nominal energy MV-CBCT with 1.4 MeV mean energy at the isocenter level was close to the performance of the 1 MeV Versa HD™ theoretical beam but did not outperform it. This is likely because, as already discussed, the component of the beam that most significantly increases DEF is its low energy portion, which gets filtered out as the beam transverses the phantom. Therefore, a shallower placement of the target increases the presence of that low energy component reaching the target. Table 13. Results of the simulation of dose enhancement over the QuAArC phantom with virtualized AuNP at 4 cm depth (“mini QuAArC configuration”), with an MLC-shaped (~2×2 cm2) field size, with Siemens ONCOR™ MV-CBCT beam, and different nominal energies of the incident beam on an Elekta Versa HD™ linear accelerator. MC error refers to the maximum absolute statistical error of the Monte Carlo simulation. Energy (MeV) DEF ± MC error Target Target center Outer ring MV-CBCT 1.14 ± 0.02 1.13 ± 0.02 1.00 ± 0.02 1 1.09 ± 0.03 1.10 ± 0.03 1.00 ± 0.03 1.5 1.06 ± 0.03 1.06 ± 0.03 1.00 ± 0.03 2 1.04 ± 0.03 1.04 ± 0.03 1.00 ± 0.03 2.5 1.03 ± 0.03 1.03 ± 0.03 1.00 ± 0.03 3 1.02 ± 0.02 1.02 ± 0.02 1.00 ± 0.02 3.5 1.02 ± 0.02 1.02 ± 0.02 1.00 ± 0.02 4 1.01 ± 0.02 1.01 ± 0.02 1.00 ± 0.02 RESULTS AND DISCUSSION 102 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy Figure 34. Representation of the dose enhancement factors obtained for MLC-shaped (~2×2cm2) fields used to irradiate the QuAArC phantom with virtualized AuNP at target depths of 10 cm (original image, blue bars) and 4 cm (“mini QuAArC configuration”, orange bars), with a Siemens ONCOR™ MVCBCT beam and different nominal energies of the incident beam on an Elekta Versa HD™ linear accelerator. The error bars represent the maximum absolute statistical error of the Monte Carlo simulation. It is worth noting that this behavior, though with a solid foundation of the physics behind it, is contrary to the one reported by Detappe et al. [36], as they reported higher sensitivity enhancement ratios for deeper tumors. Although radiosensitization is not expected to be directly proportional to dose enhancement and there could be different biological mechanisms that could explain this behavior, as it has been extensively discussed in the introduction sections 2.5 and 2.6, another explanation might be related to the energy beam they used – 6 MeV nominal energy. As illustrated in Figure 34, when nominal energies are higher, both the absolute DEF and the differences between the tested depth conditions tended to be lower. It is possible that for such energies, the described filtering effect of the phantom is negligible, and the amount of low energy particles in the spectra is mainly dependent on the particles scattered through the phantom. Hence, a deeper target would mean more chances for scattered particles to be generated along the target, and this could explain why they found an increase in the proportion of low energy photons as a function of depth in tissue. RESULTS AND DISCUSSION José Antonio López Valverde – 103 In any case, given the dependence found on the X-ray target material as the main difference for generating a softer beam in the MV-CBCT versus the theoretical Versa HT™ beams, the effect of different X-ray target materials was studied, including the conventional tungsten target (W) present in Versa HD™, a molybdenum (Mo) experimental target, and the MV-CBCT carbon-based target, as described in section 4.4.3. The results of the measured DEF with the Mo target for the “mini QuAArC” configuration can be found in Table 14. Table 14. Results of the simulation of dose enhancement over the QuAArC phantom with virtualized AuNP at 4 cm depth (“mini QuAArC configuration”), with different nominal energies of the incident beam on an Elekta Versa HD™ linear accelerator with a modified target of molybdenum. “Adjust.” stands for a field size MLC-shaped to the size of the target, approximately a 2×2 cm2 field. MC error refers to the maximum absolute statistical error of the Monte Carlo simulation. Energy (MeV) Field size (cm2) DEF ± MC error Target Target center Outer ring 1 10×10 1.23 ± 0.05 1.24 ± 0.05 1.00 ± 0.04 Adjust. 1.11 ± 0.03 1.11 ± 0.03 1.01 ± 0.03 4 10×10 1.06 ± 0.03 1.06 ± 0.03 1.00 ± 0.03 Adjust. 1.03 ± 0.02 1.03 ± 0.02 1.00 ± 0.02 It can be observed that when the Mo-based target was used, the DEF values were higher than the equivalent ones found for the W-based target, due to the lower energy on the produced spectra. Figure 35 shows the comparative results of the DEF for 4 MeV nominal energies for the W and Mo-based target, along with the MV-CBCT energy, that is close to 4 MeV (~4.2 MeV) and has a C-based target. As already predicted and discussed, lower Z materials yielded a higher DEF. Given all the conditions studied, it seems that the Siemens ONCOR™ MV-CBCT beam is the most reasonable choice when it comes to an experimental beam whose use would benefit from a higher DEF. As it was discussed in section 5.2, this beam is also a good candidate for an IGRT real-time ART implementation, which adds more value to its proposal. In addition to obtaining top results in terms of theoretical DEF and positive results as a beam for diagnostics, there is an additional factor to consider: the beam is not merely hypothetical, but is already available at HUVM, which would make it an excellent candidate for further biological experimentation. This is why this beam was proposed for studying the theoretical dose enhancement over actual RT treatment plans, along with conventional beams, as it was mentioned in section 4.4.4. RESULTS AND DISCUSSION 110 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy As discussed in previous sections, using the local dose enhancement provided by the presence of AuNP to reduce the dose to OARs, while delivering the prescribed dose to the target, would be an interesting practice. In this manner, both treatments were optimized using CARMEN-implemented optimizer mentioned in the Materials and Methods section, using fixed optimization parameters. As a result, prescribed dose was delivered to 95% of BTV on both treatments, and changes in OARs are present in the DVH (see Figure 39, bottom). Organs closer to the target receive same or slightly lower doses. However, despite these noticeable changes, both treatments are compromised by the presence of ‘hot spots’ in the healthy surrounding tissue, as seen in Figure 40. This was also quantified with CN (see Table 15), resulting in CN100 indexes of 0.311 and 0.342, with and without nanoparticles, respectively. These conditions are not desirable in clinical reality, and the optimization process should be improved, as it does not consider healthy tissue not included in OARs. Fortunately, this could be addressed in future work through some developments and modifications in our tools. Figure 40. Isodoses for evaluated plannings with MV-CBCT beam on three different CT slices for the same irradiation protocol (top images) and optimized treatments (bottom images). Thick isolines: no AuNP; thin isolines: 2.23 mg Au/g of tissue. 100% isodoses represent prescribed dose: 70 Gy. A color code is included describing structures present in the images as solid regions. RESULTS AND DISCUSSION José Antonio López Valverde – 111 Figure 41. DVHs corresponding to the percentage of each volume receiving certain dose for each RT MV-CBCT planification with 22.3 mg Au/g of tissue (dashed line) vs no AuNP (solid line). Top: same weights. Bottom: same target optimization. RESULTS AND DISCUSSION 112 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy To assess the influence of a higher AuNP concentration in dose enhancement, we studied a concentration one order of magnitude higher, with a multiplicative factor of 10. This 22.3 mg Au/g of tissue concentration provided better results. First, the average DEF was 1.159, with a standard deviation of 0.0210 (see Table 15). This is of great interest, as it gives an absorbed dose increase of ~16%, as seen in the DVH presented in Figure 41, top. This dose enhancement primarily affects the target, but similarly to the 2.23 mg Au/g of tissue simulation, some close OARs also showed a slight increase in dose. The isodoses in the planning with AuNP showed a high conformation to the target with the prescribed dose while avoiding OARs, as can be seen in Figure 42. The same treatment planning was simulated without AuNP, also shown in Figure 42. As seen, the planning with AuNP achieves an excellent result that could not be possible without the presence of nanoparticles. Figure 42. Isodoses for evaluated plannings with MV-CBCT beam on three different CT slices for the same irradiation protocol (top images) and optimized treatments (bottom images). Thick isolines: no AuNP; thin isolines: 22.3 mg Au/g of tissue. 100% isodoses represent prescribed dose: 70 Gy. A color code is included describing structures present in the images as solid regions. In Figure 42, it can be observed that, with the same irradiation protocol, treatment objectives were achieved when AuNP were present, whereas the prescription dose was scarcely delivered to the BTV when AuNP were absent. The RESULTS AND DISCUSSION José Antonio López Valverde – 113 circumscription, facilitated by the AuNP local dose enhancement, avoided the generation of ‘hot spots’ and delivered high doses with precision to the target area. CN was parametrized, resulting in CN100 values of 0.653 with nanoparticles against 0.342 without them, which would result in highly significant clinical differences. Following the schemes outlined in section 4.4.4, both treatments were optimized using the same parameters in order to see if AuNP local enhancement effect can lower the dose that the OARs receive. The resulting DVH can be observed in Figure 41, bottom. As it can be observed, prescription dose was satisfactorily delivered in both cases, while the OARs dose was reduced when AuNP were present. Another observation could be made on that same figure: the target dose heterogeneity received when AuNP were present was higher, which is directly caused by the AuNP-related local dose enhancement effects; voxels containing AuNP and surrounded by other AuNP-containing voxels receive higher doses than those voxels whose neighbors do not contain the gold-based material. Table 15. Data summary for each simulation. CN are calculated after the optimization process. Beam Energy AuNP concentration (Au mg/g of tissue) Dmax (Gy) CN100 DEF (μ ± σ) 6 MeV 0 80.5 0.389 1.009 ± 0.0152 40 81.0 0.415 MV-CBCT 0 (weights from 2.23) 79.9 0.342 1.049 ± 0.0109 2.23 83.0 0.311 0 (weights from 22.3) 70.9 0.342 1.159 ± 0.0210 22.3 78.5 0.653 Figure 43. DEF histograms of each comparative. Left: 6 MeV beam, 0 vs 40 mg Au/g of tissue. Center: MV-CBCT beam, 0 vs 2.23 mg Au/g of tissue. Right: MV-CBCT beam, 0 vs 22.3 mg Au/g of tissue In summary, the 6 MeV photons beam did not show significant differences in terms of DEF even with 40 mg Au/g of tissue (see Table 15, Figure 43), although some positive changes in isodose distribution and efficient target dose delivery, as RESULTS AND DISCUSSION 114 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy measured by the CN, could be observed (see Table 15). This is mainly related to the predominance of Compton Effect at these high energies delivered by this therapeutic beam. On the other hand, MV-CBCT, with an energy of ~1.4 MeV at the isocenter, presented a noticeable DEF even at low AuNP concentrations, such as 2.23 mg Au/g of tissue. This is the concentration found as an acceptable radiologic concentration, which resulted in an increase of ~5% in the dose received by the BTV. However, despite the lower of this beam favoring the photoelectric absorption in the BTV, the efficiency of target-dose delivery was lower, as indicated by the CN due to the formation of hotspots out of the target. Finally, increasing the radiologic concentration tenfold to 22.3 mg Au/g of tissue yielded satisfactory results, both in terms of DEF, with an increase of ~16%, and in terms of efficient target-dose delivery, achieving a CN better than that of the original treatment with 6 MeV. Nonetheless, achieving this concentration in a biological context will be more challenging than reaching lower concentrations. 5.3.2.2. Breast cancer clinical cases from the intact breast SBRT clinical trial For the breast cancer clinical cases from the intact breast SBRT clinical trial, the calculated dose enhancement due to the presence of AuNP at ~22.3 mg Au/mL was also around 15%. This dose enhancement would enable not only a simplification of the treatment reducing the number and angular complexity of the incidences, but also a better dose-volume histogram and isodoses coverage, as it can be seen in Figure 44A,B, and in the CN100 improvement (see Table 16). Notable improvements were made using AuNP treatment, both in terms of the DVH and in terms of CN. The dose coverage when AuNP were present is remarkable, with conformity numbers close to 1 thanks to the localized dose enhancement, with AuNP acting as a dose sink. Note that skin doses, unfortunately, are not directly comparable, as the commercial planner used for the 4π does not consider the particle transport in the air, not considering the dose delivered out of the body. In any case, skin dose is always difficult to calculate as a result of the voxelization involved in a so thin structure. Performing a verification of the 4π treatment with Monte Carlo for directly benchmarking both treatments calculated with the same algorithm could be an interesting work for the near future. RESULTS AND DISCUSSION José Antonio López Valverde – 115 RESULTS AND DISCUSSION 116 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy Figure 44. Sample slices with isodoses showing the coverage of the target (structure enclosed by blue lines) in the original 4π treatment and the proposed AuNP treatment, and their respective dosevolume histogram (DVH) for two independent different cases of intact tumor breast cancer (A & B). Table 16. Comparative of the treatments in terms of conformity number (CN100) Patient 4π AuNP A 0.57 0.72 B 0.33 0.95 RESULTS AND DISCUSSION José Antonio López Valverde – 117 5.3.3. Dose enhancement experimental verification Given the interesting nature of the results exposed in the previous section 5.3.2 and the need of validating those results, the experimentation proposed in the section 4.4.5 was carried out. After processing the radiochromic films, the doses measured in the films immersed in water and AuNP were 274.2 and 273.7 cGy respectively. In short, there was no difference in doses received when the films were immersed in water or in AuNP. Given these results, and the proven reliability of MC simulations made by our group, some explanations could give a hint about this experimental result. First, EBT3 film design is made of a ~28 µm layer of an active polymer which is the detector itself, and this layer is covered by a 100 µm layer of matte polyester on each side. These polyester layers could constitute a physical barrier for low energy particles, whose energies do not allow them to travel far. In this experiment, as the main component enhancing the dose are supposed to be Auger electrons, low energies are crucial, and this barrier could be hiding the enhanced radiation effect. As a reference, the electrons emitted at Auger peaks of 100 eV, 2 keV, and 10 keV have a free range in water of 50 nm, 100 nm and 1.5 µm, respectively [81]. Moreover, EBT3 films have shown significant underresponses when the radiation interacting with them presents a low energy component, presenting underresponses of ~5% to 300 keV photons, and ~20% underresponse for 70 keV photons [140]. These underresponses are caused by the active layer underresponding to these radiation energies when they reach it, it is an independent phenomenon than the one caused by the polyester barrier mentioned above. Hence, using EBT3 radiochromic films as the dosimetric detectors is not suitable for the experimentation in this project. The ideal detector would be one that could include nanoparticles right next to the detector itself, without barriers. Gel dosimetry could be considered, as it would permit including the gold nanoparticles inside the gel itself and measuring against a gel without nanoparticles. However, gel dosimetry needs an intense characterization and making the AuNP- based gel detector would be a long and inaccessible process to our research group. RESULTS AND DISCUSSION 118 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy On the other side, a refillable liquid ionization chamber, NanoLIC, was successfully designed and crafted thanks to the collaboration with the Radiophysics Lab led by Prof. Faustino Gómez Rodríguez. Same as in gel dosimetry, the AuNP solution could be the active medium itself, as already discussed in section 4.4.5, although the information would be limited to the point where the chamber is placed for each measurement, instead of a three-dimensional distribution that could be captured with gel dosimetry. However, there were several problems with the experimental measurement of DEF. First, the first prototype had several problems in terms of stability of the response, mainly caused by an unstable leakage current and problems with the refilling and sealing of the chamber. These were solved in the latest upgraded prototype. Moreover, the chamber was filled with the isooctane with purity >99.0% referenced in section 5.3.3, the reads of the chamber still showed a certain degree of unpredictability, and it still could not be zeroed. These were solved using the ultrapure isooctane, with purity >99.8%. In addition, to prevent isooctane volatilization through micro-fractures, and its consequences in leakage current drift, the chamber sealing was reinforced using Parafilm™. This allowed to stabilize the chamber readings and zero the leakage current in the high and mid-range for voltages up to ±300 V. Repeated measurements were made with 500 UM with the Elekta Versa HD™ 6 MeV FFF clinical beam from HUVM, using a 10×10 cm2 field, and the chamber positioned at the PDD maximum dose region, approximately. These measurements were stable, and yielded measurements around 355 nC, with a detected precision up to the unit. As for the moment of writing this work, NanoLIC is ready for testing and measuring. However, the MV-CBCT beam is no longer available since ONCOR™ linear accelerator was decommissioned recently. This experiment for validating DEF will be resumed soon since we now have another MV-CBCT beam on a Varian linac model installed at HUVR. However, its characterization and Monte Carlo modeling have not been carried out yet, as we do not yet have the necessary experimental measurements at our disposal. RESULTS AND DISCUSSION José Antonio López Valverde – 119 5.4. Gold nanoparticles as radiosensitizers 5.4.1. External beam radiotherapy Images of the γH2AX foci immunofluorescence were acquired successfully using the STED super-resolution technique. Some representative fragments of the tile scans obtained are shown in Figure 45. A trend of higher foci density can be observed when the doses increase, as expected. Despite a higher γH2AX foci density can be slightly observed naked eye when nanoparticles are present, especially at higher doses, the differences between control cells and AuNP-treated cells are not clear without a digital quantification process of the images. Quantification using the SoFoCo algorithm confirmed the relationship between higher γH2AX foci density and higher radiation doses, as can be seen in Figure 46. Linear regression for both datasets yielded R2 values 0.9847 and 0.9977 for the control and AuNP data, respectively. Both correlations were statistically significant with a p-value < 0.001. Given that the last incubation with DMEM or AuNP, for control and AuNP datasets, respectively, changes the culture conditions, the data in Figure 46A,B are not directly comparable. However, analyzing the increase in relative foci density between each group using the slopes of the linear regression equation, it can be seen that the DNA damage density increases more significantly with nanoparticles, with a slope of ~3.25 (see Figure 46B), than without nanoparticles, with a slope of ~2.23 (see Figure 46A). RESULTS AND DISCUSSION 126 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy CBCT 10x10 cm2 phase space and scoring ~1.8·108 histories and ~3·108 particles in each well for the second phase of the simulation. At the second stage of the simulation, the dose at the nucleus and the dose at the cell was scored, and the results are shown in Table 17. Table 17. Comparative of the dose at cell and nucleus for each simulation scenario Simulation Dose at cell (Gy) Dose enhancement Dose at nucleus (Gy) Dose enhancement Control 0.0136008 - 0.0134368 - AuNP, 0 mg Au/mL internalized 0.0144763 1.0644 0.0143214 1.066 AuNP, 2.23 mg Au/mL internalized 0.0144739 1.0642 0.0143322 1.067 AuNP, 11.15 mg Au/mL internalized 0.0144669 1.0637 0.0143076 1.065 AuNP, 22.2 mg Au/mL internalized 0.0146620 1.0780 0.0142631 1.061 As predicted, those results show a consistent dose enhancement at the cellular level when nanoparticles are present. However, the magnitude of the dose enhancement is lower than the one reported by the macroscale simulations. This could be caused by different simplifications made in the model, such as the impossibility of considering the phantom output factor energy contribution as the cells were irradiated from below and the second phase space was scored before there was any interaction with the media in the wells. Nonetheless, the dose enhancement variation depending on the internalization ratio showed a sawtooth pattern. This could mean that the statistics of the second phase of the simulation are low and more demanding simulations should be performed, or that there are some mistakes on the code interpretation and parametrization of the distribution of AuNP. Further work in the model is needed to assess this matter, and we are closely monitoring this with the TOPAS-nBio developer team. Anyhow, the mere formulation of this multiscale model, its geometrical definition and the connection between scales can already be considered as a success. The inputs and geometries designed for the third stage of the simulation were not simulated because of their high amount of computational resources consumption RESULTS AND DISCUSSION José Antonio López Valverde – 127 and the reported uncertainties in the second stage. This third stage, once completed, should allow direct comparison, refining, and benchmarking with experimental measurements. 5.5. Ethics committee for animal experiments The biological experimentation foundations for works in progress were settled satisfactorily. The corresponding ethical committee gave approval to the proposed biological experimentation, both for in vitro studies using primary cell cultures derived from rat liver, and in vivo studies for assessing AuNP pharmacokinetic parameters and toxicity. To design and accomplish those biological experiments, proper training was received, leading the author of this thesis to obtain the legal required capacitance for manipulating experimental animal models – functions A, B and C in accordance to fulfil Spanish and European laws. Also, function D required to design proceedings involving animal models was obtained by the principal investigator of our group. The formal acceptance has been received from IFAPA, the Andalusian institution which allows us to begin the biologic methodology included in this project for which the results presented here are the essential starting point. As an additional result of this work, we include here (see Figure 50) the approval from the Ethics Committee required for the achieved lawful acceptance. Figure 50. Ethics Committee letter of approval. RESULTS AND DISCUSSION 128 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy CONCLUSIONS José Antonio López Valverde – 129 6. Conclusions Considering the obtained results, the following conclusions can be elicited: 1. A robust and reproducible method for synthesizing AuNP was implemented. Although these nanoparticles were used for physics experimentation, further chemical work is needed for testing and increasing their biocompatibility. 2. AuNP contrast in a clinical context could be observed at concentrations around 2 mg Au/mL. Moreover, the established AuNP concentration value of around 20 mg Au·mL−1 could be used both for diagnosis and for its application in treatments, under the conditions of a Cone Beam CT based on megavoltage beam for IGRT. To the extent that AuNP are conveniently retained by the tumor mass, IGRT would provide a functional image beyond the usual one to correct patient positioning, which would mean monitoring the course of the disease throughout the radiotherapy treatment sessions. 3. Considering the influence of key parameters on dose enhancement, the MVCBCT beam seems to be adequate because of its nominal energy and spectrum due to the presence of the carbon target. It presents a spectrum component hard enough for being able to penetrate through the patient, and at the same time, soft enough to promote dose enhancement, especially for cases with tumors not too deep, such as the typical ones in the head and neck and others, such as in breast. 4. The dose enhancement shown could be implemented in treatment planning optimization for relative dose distribution calculation and increasing the dose in the PTV while decreasing the dose in OAR. Furthermore, this dose enhancement would be delivered at the cell level, so AuNP implementation is part of the role of the precision medicine in radiotherapy. 5. The proposed AuNP concentration has shown therapeutic efficacy as radiotherapy dose enhancer with the MV-CBCT beam for IGRT. In these conditions, dose enhancement was increased to ~15% based on in silico experiments carried out with accurate modeling of the physical problem. 6. Furthermore, in vitro experiments based on γH2AX foci methodology and micronuclei assay showed even more exciting results on radiosensitization by achieving a ~47% improvement. CONCLUSIONS 130 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy 7. These results are relevant insofar as they were achieved under clinical conditions and beam energy configuration that could even provide a treatment monitoring procedure. These results now encourage an in vivo evaluation on a physiologically and dimensionally convenient animal model, having notably reduced the number of sacrifices that would have been necessary to achieve the results of this work without following a sequential integrated strategy. 8. Gold nanoparticles dosimetric experimentation requires more exigent conditions than conventional radiotherapy treatments dosimetric validation experiments. Due to the physical characteristics involved in dose enhancement, using radiochromic films is discarded, and other experimental setups will be needed to experimentally assess gold nanoparticles dose enhancement capacity, such as liquid chamber dosimetry with nanoparticles inside. 9. The basis of a multi-scale radiosensitization model were settled, priming future developments and advancements in this direction. 10. The next steps of in vivo experimentation were outlined and the ethics committee approval for these experiments was granted, ensuring the next steps towards the clinical implementation of gold nanoparticles. REFERENCES José Antonio López Valverde – 131 7. References 1. Instituto Nacional de Estadística (INE). (2023). Defunciones según la Causa de Muerte - Primer semestre 2023 (datos provisionales) y año 2022 (datos definitivos). Retrieved from https://www.ine.es/prensa/edcm_2022_d.pdf 2. Ferlay, J., Ervik, M., Lam, F., Laversanne, M., Colombet, M., Mery, L., … Bray, F. (2024). Global Cancer Observatory: Cancer Today. Lyon, France. Retrieved from https://gco.iarc.fr/today 3. Sung, H., Ferlay, J., Siegel, R. L., Laversanne, M., Soerjomataram, I., Jemal, A., & Bray, F. (2021). Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA: a cancer journal for clinicians, 71(3), 209–249. https://doi.org/10.3322/CAAC.21660 4. Miller, K. D., Siegel, R. L., Lin, C. C., Mariotto, A. B., Kramer, J. L., Rowland, J. H., … Jemal, A. (2016). Cancer treatment and survivorship statistics, 2016. CA: A Cancer Journal for Clinicians, 66(4), 271–289. https://doi.org/10.3322/caac.21349 5. Atun, R., Jaffray, D. A., Barton, M. B., Bray, F., Baumann, M., Vikram, B., … Gospodarowicz, M. (2015). Expanding global access to radiotherapy. The Lancet. Oncology, 16(10), 1153–86. https://doi.org/10.1016/S1470-2045(15)00222-3 6. Barbeiro, A. R., Ureba, A., Baeza, J. A., Linares, R., Perucha, M., Jiménez- Ortega, E., … Leal, A. (2016). 3D VMAT Verification Based on Monte Carlo Log File Simulation with Experimental Feedback from Film Dosimetry. PLOS ONE, 11(11), e0166767. https://doi.org/10.1371/journal.pone.0166767 7. Charaghvandi, K. R., van’t Westeinde, T., Yoo, S., Houweling, A. C., Rodrigues, A., Verkooijen, H. M., … van den Bongard, H. J. G. D. (2019). Single dose partial breast irradiation using an MRI linear accelerator in the supine and prone treatment position. Clinical and Translational REFERENCES 132 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy Radiation Oncology, 14, 1–7. https://doi.org/10.1016/j.ctro.2018.09.001 8. Charaghvandi, R. K., van Asselen, B., Philippens, M. E. P., Verkooijen, H. M., van Gils, C. H., van Diest, P. J., … van den Bongard, H. J. G. D. (2017). Redefining radiotherapy for early-stage breast cancer with single dose ablative treatment: a study protocol. BMC Cancer 2017 17:1, 17(1), 1–9. https://doi.org/10.1186/S12885-017-3144-5 9. Bentzen, S. M. (2005). Theragnostic imaging for radiation oncology: dose-painting by numbers. The Lancet Oncology, 6(2), 112–117. https://doi.org/10.1016/S1470-2045(05)01737-7 10. Jiménez-Ortega, E., Ureba, A., Vargas, A., Baeza, J. A., Wals-Zurita, A., García-Gómez, F. J., … Leal, A. (2017). Dose painting by means of Monte Carlo treatment planning at the voxel level. Physica Medica, 42, 339– 344. https://doi.org/10.1016/j.ejmp.2017.04.005 11. Ling, C. C., Humm, J., Larson, S., Amols, H., Fuks, Z., Leibel, S., & Koutcher, J. A. (2000). Towards multidimensional radiotherapy (MD-CRT): biological imaging and biological conformality. International journal of radiation oncology, biology, physics, 47(3), 551–60. https://doi.org/10.1016/s0360-3016(00)00467-3 12. Foster, B., Bagci, U., Xu, Z., Dey, B., Luna, B., Bishai, W., … Jain, S. (2014). Segmentation of PET Images for Computer-Aided Functional Quantification of Tuberculosis in Small Animal Models. IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, 61(3). https://doi.org/10.1109/TBME.2013.2288258 13. Kong, F.-M., Ten Haken, R. K., Schipper, M., Frey, K. A., Hayman, J., Gross, M., … Kalemkerian, G. P. (2017). Effect of Midtreatment PET/CT- Adapted Radiation Therapy With Concurrent Chemotherapy in Patients With Locally Advanced Non–Small-Cell Lung Cancer. JAMA Oncology, 3(10), 1358. https://doi.org/10.1001/jamaoncol.2017.0982 14. Berwouts, D., Olteanu, L. A. M., Duprez, F., Vercauteren, T., De Gersem, W., De Neve, W., … Madani, I. (2013). Three-phase adaptive dose- REFERENCES José Antonio López Valverde – 133 painting-by-numbers for head-and-neck cancer: initial results of the phase I clinical trial. Radiotherapy and Oncology, 107(3), 310–316. https://doi.org/10.1016/J.RADONC.2013.04.002 15. Meijer, G., Steenhuijsen, J., Bal, M., De Jaeger, K., Schuring, D., & Theuws, J. (2011). Dose painting by contours versus dose painting by numbers for stage II/III lung cancer: Practical implications of using a broad or sharp brush. Radiotherapy and Oncology, 100(3), 396–401. https://doi.org/10.1016/j.radonc.2011.08.048 16. Duprez, F., De Neve, W., De Gersem, W., Coghe, M., & Madani, I. (2011). Adaptive Dose Painting by Numbers for Head-and-Neck Cancer. International Journal of Radiation Oncology*Biology*Physics, 80(4), 1045–1055. https://doi.org/10.1016/j.ijrobp.2010.03.028 17. Grau, C., Overgaard, J., Høyer, M., Tanderup, K., Lindegaard, J. C., & Muren, L. P. (2015). Biology-guided adaptive radiotherapy (BiGART) is progressing towards clinical reality. Acta Oncologica, 54(9), 1245– 1250. https://doi.org/10.3109/0284186X.2015.1076992 18. Thorwarth, D. (2018). Biologically adapted radiation therapy. Zeitschrift für Medizinische Physik, 28(3), 177–183. https://doi.org/10.1016/j.zemedi.2017.08.001 19. Paterson, C., Hargreaves, S., & Rumley, C. N. (2020). Functional Imaging to Predict Treatment Response in Head and Neck Cancer: How Close are We to Biologically Adaptive Radiotherapy? Clinical Oncology, 32(12), 861–873. https://doi.org/10.1016/j.clon.2020.10.004 20. Cysouw, M. C. F., Kramer, G. M., Hoekstra, O. S., Frings, V., de Langen, A. J., Smit, E. F., … Boellaard, R. (2016). Accuracy and Precision of Partial- Volume Correction in Oncological PET/CT Studies. Journal of Nuclear Medicine, 57(10), 1642–1649. https://doi.org/10.2967/jnumed.116.173831 21. van der Vos, C. S., Koopman, D., Rijnsdorp, S., Arends, A. J., Boellaard, R., van Dalen, J. A., … Visser, E. P. (2017). Quantification, improvement, and harmonization of small lesion detection with state-of-the-art PET. REFERENCES 134 – Feasibility Evaluation of the Use of Gold Nanoparticles as Theragnostic Agents for Personalized and Precision Radiotherapy European Journal of Nuclear Medicine and Molecular Imaging, 44(S1), 4–16. https://doi.org/10.1007/s00259-017-3727-z 22. Bai, B., Bading, J., & Conti, P. S. (2013). Tumor quantification in clinical positron emission tomography. Theranostics, 3(10), 787–801. https://doi.org/10.7150/thno.5629 23. Balcerzyk, M., Fernández-López, R., Parrado-Gallego, Á., Pachón- Garrudo, V. M., Chavero -Royan, J., Hevilla, J., … Leal, A. (2017). Application of EARL (ResEARch 4 Life®) protocols for [18F]FDGPET/CT clinical and research studies. A roadmap towards exact recovery coefficient. Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, 873, 39–42. https://doi.org/10.1016/j.nima.2017.04.035 24. Jiménez-Ortega, E., Agüera, R., Ureba, A., Balcerzyk, M., Wals-Zurita, A., García-Gómez, F. J., & Leal, A. (2022). Implications of the Harmonization of [18F]FDG-PET/CT Imaging for Response Assessment of Treatment in Radiotherapy Planning. Tomography, 8(2), 1097–1112. https://doi.org/10.3390/tomography8020090 25. Baeza, J. A., Ureba, A., Jimenez-Ortega, E., Barbeiro, A. R., Lagares, J. I., & Plaza, A. L. (2015). SU-E-T-157: CARMEN: A MatLab-Based Research Platform for Monte Carlo Treatment Planning (MCTP) and Customized System for Planning Evaluation. Medical Physics, 42(6Part14), 3367– 3368. https://doi.org/10.1118/1.4924519 26. Medical Physics Research Group, Medical Physiology and Biophysics Department, U. of S. (n.d.). CARMEN - Monte Carlo Radiotherapy Treatment Planning for Research Purposes. Retrieved February 2, 2024, from http://grupos.us.es/medicalphysics/ 27. Ureba, A., Salguero, F. J., Barbeiro, A. R., Jimenez‐Ortega, E., Baeza, J. A., Miras, H., … Leal, A. (2014). MCTP system model based on linear programming optimization of apertures obtained from sequencing REFERENCES José Antonio López Valverde – 135 patient image data maps. Medical Physics, 41(8Part1). https://doi.org/10.1118/1.4890602 28. Ferrari, M. (2005). Cancer nanotechnology: opportunities and challenges. Nature Reviews Cancer, 5(3), 161–171. https://doi.org/10.1038/nrc1566 29. Tietze, R., Lyer, S., Dürr, S., & Alexiou, C. (2012). Nanoparticles for cancer therapy using magnetic forces. Nanomedicine, 7(3), 447–457. https://doi.org/10.2217/nnm.12.10 30. Afzal, O., Altamimi, A. S. A., Nadeem, M. S., Alzarea, S. I., Almalki, W. H., Tariq, A., … Kazmi, I. (2022). Nanoparticles in Drug Delivery: From History to Therapeutic Applications. Nanomaterials, 12(24), 4494. https://doi.org/10.3390/nano12244494 31. Liu, Y., Ai, K., & Lu, L. (2012). Nanoparticulate X-ray Computed Tomography Contrast Agents: From Design Validation to in Vivo Applications. Accounts of Chemical Research, 45(10), 1817–1827. https://doi.org/10.1021/ar300150c 32. Cole, L. E., Ross, R. D., Tilley, J. M. R., Vargo-Gogola, T., & Roeder, R. K. (2015). Gold nanoparticles as contrast agents in x-ray imaging and computed tomography. Nanomedicine, 10(2), 321–341. https://doi.org/10.2217/nnm.14.171 33. Lee, N., Choi, S. H., & Hyeon, T. (2013). Nano‐Sized CT Contrast Agents. Advanced Materials, 25(19), 2641–2660. https://doi.org/10.1002/adma.201300081 34. Hainfeld, J. F., Slatkin, D. N., Focella, T. M., & Smilowitz, H. M. (2006). Gold nanoparticles: a new X-ray contrast agent. The British Journal of Radiology, 79(939), 248–253. https://doi.org/10.1259/bjr/13169882 35. Hainfeld, J. F., O’Connor, M. J., Dilmanian, F. A., Slatkin, D. N., Adams, D. J., & Smilowitz, H. M. (2011). Micro-CT enables microlocalisation and quantification of Her2-targeted gold nanoparticles within tumour regions. The British journal of radiology, 84(1002), 526–33. https://doi.org/10.1259/bjr/42612922