Radiation nanosensitizers in cancer therapy—From preclinical discoveries to the outcomes of early clinical trials

Abstract Improving the efficacy and spatial targeting of radiation therapy while sparing surrounding normal tissues has been a guiding principle for its use in cancer therapy. Nanotechnologies have shown considerable growth in terms of innovation and the development of new therapeutic approaches, particularly as radiosensitizers. The aim of this study was to systematically review how nanoparticles (NPs) are used to enhance the radiotherapeutic effect, including preclinical and clinical studies. Clinicaltrials.gov was used to perform the search using the following terms: radiation, cancer, and NPs. In this review, we describe the various designs of nano‐radioenhancers, the rationale for using such technology, as well as their chemical and biological effects. Human trials are then discussed with an emphasis on their design and detailed clinical outcomes.


| HISTORY OF RADIATION IN CANCER
The first medical X-ray image of a hand is attributed to German physicist Wilhelm Conrad Röntgen in 1895. In the years to follow, it became clear that the use of X-rays was associated with side effects such as skin burns 1 and hair loss. 2 Thus, the idea of using them as a potential mean of treating tumors emerged. In parallel, there have been many advances in the development of X-ray tubes. 3 X-rays are produced from electrical current, that is, an electron beam discharged from a hot cathode in vacuum. The beam of electrons produces X-rays as it encounters the anode or glass wall (in the case of the earlydeveloped X-ray tubes, e.g., the Coolidge tube, in 1913). The first reported case of using radiation therapy to treat cancer with a similar device was reported in 1896 by French Physician Dr. Victor Despeignes for the exploratory treatment of a patient with gastric cancer. 4,5 Then, the first instance of brachytherapy shortly followed Drs. Pierre and Marie Curie's discovery of radium in 1898. In July 1903, Dr. Alexander Graham Bell suggested the use of brachytherapy in a letter to Dr. Z. T. Sowers; the correspondence was published in the journal Nature. 6 In October 1903, Dr. Margaret Abigail Cleaves, an American physician focused on psychological and gynecological disorders, was the first to use radium in the treatment of gynecologic malignancies via brachytherapy. 7 Since then, the management of cancer patients using radiation has significantly evolved, with the greater understanding of the physics and adverse events (AEs) that accompany X-ray therapy. Indeed, radiation-induced leukemia or malignant skin changes were evidenced as early as 1900. 8 Optimization of the dose intensity delivered and its fractionation have been some of the major improvements, along with the deployment of modalities that have significantly improved the spatial targeting of radiation. 9,10 Over the last few years, conventional radiotherapy has progressively been replaced by conformal radiotherapy (CFRT) where the radiation beam is geometrically controlled to fit the tumor shape while sparing the surrounding organs. Additionally, intensity-modulated radiation therapy (IMRT) has become the gold standard in radiotherapy treatment by allowing a controlled irradiation field shape, like conventionally fractionated radiation therapy (CFRT), but its intensity is also modulated within the irradiated area. Today, IMRT is associated with image-guided radiotherapy and radiotherapy, in general, is now a common component of multidisciplinary cancer care, used in conjunction with surgery and other systemic therapies. 9 More recently, a new irradiation modality called hadron therapy has been developed, using charged particles instead of photons. 10 In 1946, Robert R. Wilson was the first to propose the use of proton beams for the treatment of cancer. 11 The major advantage of this technique is its depth-dose profile characterized by a significant increase in the dose deposited at the end of the particle path. 12 Based on clinical data and in vivo experiments, Paganetti et al. have demonstrated that the relative biological effectiveness (RBE) of protons was close to 1.1. This implies that protons lead to the same cancer cell death in comparison to photons, with a 10% reduced dose of radiation delivered to tissues. 13 RBE depends on several parameters including the linear energy transfer (LET), which is why the RBE of charged particles varies along their path and is significantly higher near the Bragg peak. Photons have a low LET coefficient, meaning that they ionize atoms in the tissue that are spaced by several tenths of a micrometer apart, sparsely and randomly along their path. In contrast, LET is higher for protons, which generate more radicals per particle track than lower-LET ones. 14 Finally, hadron therapy relies on two key parameters 15 : (i) ballistic features allowing a dose optimization into tumor volume, sparing surrounding healthy tissues, and (ii) biological features that allow a greater RBE associated with a high LET along their path. Using this method, the dose can be deposited in a chosen area with a high accuracy. Proton therapy is an especially good option for the radiation therapy of pediatric cancers, given the sensitivity of developing organs to radiation and potential long-term side effects. 16

| INTERACTION OF PHOTONS WITH MATTER
The photon attenuation in matter is due to three different processes: photoelectric effect, Compton scattering (Rayleigh scattering can be neglected), and pair production. 17 A photon can transfer its energy to one electron of the target leading to an electron ejection from the atom, that is, the photoelectric effect. During this effect, the energy of the incident photon is totally transferred to an electron in an inner shell (photoelectron). The vacancy created in the inner layer is filled by an electron from an outer layer, the energy being released in the form of a fluorescence X-ray photon or an Auger electron. For K shell vacancies, the Auger yield decreases with atomic number (Z), and for Z = 30 (zinc) the probabilities of the emission of X rays from the innermost shell and of the emission of Auger electrons is about equal. 18 When Compton scattering occurs, the electron is spread away together with a new photon that has a lower energy than the incoming one. Highly energetic photons (E > 1.02 MeV) produce an electron-positron pair. This positron slows down in matter and strikes with an electron leading to the formation of two gamma rays of 0.511 MeV. Photons are produced in opposite directions. Depending on the atomic number of the material, each phenomenon contributes in a different way to the total photon attenuation. At low energy (below 100 keV), the photoelectric effect governs the attenuation. For energies between 100 keV and 10 MeV, the Compton effect becomes the primary process for attenuation, regardless of the intervening material. 19  Tumor growth is associated with angiogenesis, the formation of a vascular bed surrounding the tumor, in order to provide it with essential nutrients that support continued cancer growth. The newly formed vasculature is in essence disorganized and leaky, which provides an ideal environment for nanosized particles injected in the bloodstream to passively permeate into tumor tissue. This effect has been named the enhanced permeability and retention (EPR) effect. 20 Nanoparticles (NPs) end up at the tumor site due to the porosity of the local endothelium and are subsequently retained in the tumor because of the ineffective lymphatic drainage at the tumor level. 20 Thus, NPs represent an ideal candidate to deliver chemotherapeutic drugs to tumor site, minimizing side effects to healthy tissues. Even though some nanomedicines have reduced toxicity for patients compared to free drug (e.g., Doxil vs. Doxorubicin), their accumulation in the tumor usually represents a very limited fraction of the injected dose. 21 Indeed, Wilhelm et al. reviewed 10 years of data and concluded that 0.7% of the injected dose typically reaches the tumor site. 21 Thus, strategies have been developed for particles to actively target the tumor site by tagging them with molecules able to recognize ligands within the peripheral vascular bed (e.g., vascular endothelial growth factor (VEGF) receptor 22 ), the extracellular matrix 23 or on the surface of cancer cells (e.g., folate receptor, 24 epidermal growth factor receptor (EGFR), human epidermal growth factor receptor 2 (HER2) 25,26 ). Interestingly, tumor homing can also be enhanced by utilizing tubularshaped NPs, 25 which undergo high levels of phagocytosis by immune cells compared to their spherical counterparts. These cells subsequently travel to the tumor via the bloodstream, delivering the nanotubes and their drug payloads. Additionally, tagging NP surface with the marker of self (CD47) delays their recognition and clearance, thus increasing their ability to accumulate at tumor site. 26 Similar strategies based on coating the NPs with RBC (red blood cell) 27 or leukocyte 28 membranes have successfully extended the circulation time of NPs in murine models. Using external stimuli to improve targeting has also been explored, such as using ultrasound stimulation to break nanoassembled microparticles, exclusively at tumor site, thus enhancing their accumulation. 29 Targeted delivery using an external magnetic field is also possible; iron oxide NPs represent an efficient means of drug delivery in that case. 30 Hyperthermia or ultrasound have also been used to break biological barriers at microscopic level (e.g., blood-brain barrier [BBB]) and improve nanomedicine delivery 30 (Table 1).
Besides utilizing active targeting, local administration of NPs can allow for increased accumulation of the radiosensitizers at the tumor site, while limiting the potential toxicity associated with systemic circulation. 31 Hence, the recent review by Boateng and Ngwa 31 highlights the various possible routes of administration of nanoradioenhancers from passive delivery systems to implantable-sustained release systems. Utilizing "smart" spacers (i.e., implantable release systems) during radiotherapy allows for NP radioenhancers to be locally applied. 32 As spacers are often used during radiotherapy to guide the geometric localization of the treatment, switching from inert spacers to "smart" spacers would not require additional procedures. 32 Inhalation of radiosensitizers also allows for local administration to the lungs and has been shown to be effective. 33 Lastly, as the liver and spleen are the primary clearance organs of such biomaterials, strategies consisting of priming/saturating the liver with ghost particles before administering the nanomedicine have been successfully achieved. 34 The parameters affecting biodistribution, including circulation time following IV injection of NPs, are highly dependent on NP's intrinsic properties such as their size, composition, surface charge, surface functionalization (along with grafting density), shape, etc. 35 In addition, NP interaction with plasma proteins leads to the formation of an adsorbed protein corona that further modulates NP fate as it can affect its resulting size, surface charge, and overall stability. 35,36 PEGylation of the NP surface is a widely used strategy to significantly decrease plasma protein adsorption and improve circulation time of NPs. 37 Nano-radioenhancers are capable of enhancing the sensitivity of cancer cells to radiation. The biological effects subsequent to the use of nano-enhancers to potentiate tumor cell radiation have been reviewed by Sun et al. 38 and include: (i) reactive oxygen species (ROS) production (leading to oxidative stress, also called the chemical phase 39 ), as well as (ii) DNA damage. Both the levels of ROS production and DNA damage appear to be inversely proportional to the size of NPs, 38 suggesting that particles with larger surface area to volume ratios produce more ROS and DNA damage. Next, nano-sensitizerassisted radiation also induces (iii) cell cycle arrest in the G2/M transition, [40][41][42] which then leads to apoptosis ( Figure 1). However, cell cycle arrest in G1/S followed by senescence has also been observed with nanodiamonds. 43 Mitochondrial involvement has been highlighted in multiple studies. 44,45 For example, a study by Ghita et al. used soft X-ray microbeam (carbon K-shell, 278 eV) to achieve subcellular targeting of radiation (i.e., cytoplasmic and nuclear irradiation) in an effort to better understand the mechanistic effects of gold nano-radioenhancers. 44 Exclusive cytoplasmic irradiation of MDA-MB-231 human breast cancer cells combined with 1.9 nm Aurovist™ (located in the cytoplasm) still led to DNA damage, along with mitochondrial depolarization (oxidation). Ghita et al. highlighted that modulation of the physico-chemical parameters of the particles might lead to different effects. 44 Many studies describe the NP's ability to enhance cell death upon irradiation. Interestingly, some papers report significant radiosensitization effects in vitro with very low quantity of NPs. 46 Regarding this study, 46 Penninckx et al. calculated that 0.001% of gold NPs per mass (650 NPs per HepG2 human liver cancer cells) led to a radiosentizing effect which was 250 times greater than predicted. 39 This highlights that critical factors for radioenhancement go beyond the dose of NPs. 47  and intrinsic radiosensitivity (tumor and/or patient-specific response).
As previously mentioned with several examples, NPs may synergize with several of these factors. In a recent review, Penninckx et al. summarize how gold NPs, used as radioenhancers, affect these 5-R factors at the molecular and cellular levels. 39 In this review, the authors also point out the major differences induced in these factors for low LET radiation (mainly X-rays) and high LET radiation (i.e., protons, alpha rays, or heavy ions) when combined with nano-radioenhancers. In particular, they notice that, independent of the proton energy or the gold NP size/concentration, the physical enhancement is negligible, even if significant radiosensitization effects are observed. 53

| NPs with intrinsic radioenhancer properties
The ability of NPs to act as X-ray radiosensitizers is commonly explained via the physical phenomena described in Section 2, and particularly due to the increased absorption of X-rays associated with the emission of secondary electrons and fluorescence photons. 56 These phenomena lead to an enhancement of energy deposition. Thus, in the context of radio-enhancement, priority has been given to high atomic number elements. The combination of photon radiation and heavy NPs leads to local radiation hardening and higher LET. 57 Among them, gold (Z = 79) has been widely studied for radiation therapy due to its biocompatibility. 58 The gadolinium doubles as a contrast agent along with a radiosensitizer. 66  that radiation-exposed hydrogenated nanodiamonds displayed significantly higher ROS compared to both radiation and particle effect alone, as well as DNA damage, cell cycle arrest, and senescence. 43 Mirjolet

| NPs used to deliver radiation-enhancer molecules
NPs can also be used as a vehicle to deliver radiosensitizing drugs or molecules to the tumor site rather than the NPs themselves sensitizing the tumor. One common strategy is to deliver chemotherapy to the tumor which can enhance the effects of radiation treatment. Cisplatin, a commonly used chemotherapy, enhances the effects of radiation through its interaction (i.e., high reactivity and electron-transfer reactions) with the electrons generated during radiation. 74 Liposomal cisplatin has been found to be effective both in vitro and in vivo against Lewis lung carcinoma. Furthermore, it was found to be more effective as a radiosensitizer than free cisplatin due to increased accumulation within cancer cells. 75 Cisplatin has also been conjugated to gold NPs for tumor sensitization of head and neck cancers, as well as glioblastoma. 76 also been used to treat lung and pancreatic cancer in vitro, using Dalpha-tocopheryl PEG 1000 to assist with cellular uptake. 83 Docetaxel has also been loaded onto titanate nanotubes to treat prostate cancer, 70 as well as a PCL (polycaprolactone) NP for gastric cancer. 84 Other chemotherapeutics delivered in nano-formulations for radiosensitization include paclitaxel, 85 doxorubicin, 86 and topotecan. 87 Another common strategy was to deliver molecules that reduced the effects of hypoxia-related resistance to radiotherapy.
Nitroimidazole is an imaging agent specific to hypoxia; it manages to sensitize tumors to radiation by generating ROS. 88 Researchers have made "smart" nanogels loaded with IAZA (iodoazomycin arabinoside), a nitroimidazole derivative, and functionalized with galactose. These nanogels were able to sensitize hepatocellular carcinoma (HCC) cells under hypoxic conditions in vitro. 88

| Active targeting of nanosized radioenhancers
In an effort to further improve the targeting efficiency, nanosized radioenhancers have been functionalized with moieties able to actively target the tumor, its microenvironment, or the associated vasculature. One common method utilized is by conjugating antibodies to the particle surface. EGFR is a receptor whose over-expression in cancer cells is linked to cell proliferation, angiogenesis, and tumor metastasis. 93 One study functionalized gold NPs with anti-EGFR antibodies, where they were able to sensitize the effects of proton irradiation in cells over-expressing EGFR but not in cells lacking EGFR. 94 Another study used anti-EGFR antibody functionalized to gold NPs to deliver β-lapachone, an anticancer agent. 95 These NPs preferentially accumulated in cancer cells according to the amount of EGFR expressed, with higher accumulation occurring in A431s than in A549s, though both had more accumulation in comparison to RKO cells, which lack EGFR.
They successfully radio-sensitized tumors following IV injection in a mouse model with xenografted A549 tumors. 95 Similarly, both ironoxide NPs and silver NPs also have been functionalized to increase sensitivity to radiation for radioresistant glioblastoma and nasopharyngeal carcinoma cells, respectively. 96,97 In addition, HER2, overexpressed in some cancers, has also been leveraged to improve tumor targeting/treatment of breast, pancreatic, ovarian, endometrial, gastric, and esophageal cancers. 98  with poor clinical outcome. 100 Anti-HER2 functionalized gold and silver NPs have also been used to radiosensitize breast cancer. 101,102 Other antibodies used for radioenhancer NPs include Anti-RhoJ, which is expressed in the vasculature of peri-and intratumoral regions, and cmHsp70.1 antibody, which targets a heat shock protein expressed on aggressive glioma cells. [103][104][105] Another commonly utilized strategy for targeting is to conjugate the particles with folate or folic acid. One study comparing the efficacy of nano-radiosensitizers decorated with folic acid, glucose, or glutamine found that both glutamine and folic acid significantly increase the efficacy of the radiosensitizers for breast cancer. However, neither showed significant advantage over the other. 106 Despite this, using folic acid and folate remains a major strategy for tumor targeting nano-radioenhancers. Combined folate-and RBC membrane-functionalized bismuth NPs enabled an increased survival in mouse models of breast cancer compared to the nontargeted NPs and radiation alone. 107 Other studies have used bovine serum albumin NPs with folate to target breast cancer in vitro. 108 Multiple studies have developed nano-radioenhancer formulations utilizing folate targeting nasopharyngeal cancer as a model of head and neck cancers. All of these studies demonstrated radio-sensitizing efficacy using KB cells, which overexpress the folate receptor. 82,109,110 In vitro studies done by Shakeri-Zadeh et al. used folate conjugated gold NPs and nanorods to enhance the effects of radiation and photothermal therapy. 109,110 Werner et al. found that, in vivo, PLGA-lecithin-PEG NPs containing docetaxel with folate were more effective at radio-sensitizing KB cell tumors than free docetaxel or the nontargeted NPs. 82 Folate-conjugated NPs have also been shown to target gliomas for radio-sensitization, as folate receptors are overexpressed in some brain tumors, as well as on the luminal side of the BBB endothelial cells, which can help bring folate-conjugated NPs into the brain through the BBB. 111,112 A variety of other strategies have also been utilized. Conjugating particles with Arg-Gly-Asp (RGD) has been used for the radiosensitization of lung, breast, and cervical cancer. RGD peptides recognize a few integrins, including the α v β 3 integrin, which have increased expression on tumor blood vessels and some cancer cells. [113][114][115][116][117][118] Thio-glucose is another targeting modality used for a variety of different cancers. Cancer cells exhibit a higher glucose metabolism than normal tissues, resulting in preferential uptake of the thio-glucose bound NPs than by normal tissue cells. [119][120][121] In addition, glioma cells, along with the BBB, express low-density lipoprotein receptor-related protein-1 (LRP-1) which can be targeted with angiopep-2 conjugated NPs for radio-enhancement. 89,122 It is important to note that the majority of cancer nanomedicines that are approved or undergoing clinical trials rely on passive targeting. 123 In addition, active targeting strategies do not fully address the issue of off target effects 123 as the targeted receptors are also expressed on normal tissue, though to a lesser extent (e.g., EGFR, HER2, transferrin, folate receptors, etc.). 124

| Radiosensitizer combination for an enhanced effect
Radiosensitizer combinations have also been explored to achieve an enhanced therapeutic effect, including, but not limited to, tandems of: (i) two NP radiosensitizers, 125 (ii) NP radiosensitizers and a chemotherapeutic drug, 70 (iii) NP radiosensitizers and tumor oxygenation 126 or including (iv) dual effect NPs displaying radiosensitizing and glutathione trapping effect. 127 Indeed, Cheng et al. engineered dumbbelllike NPs made of gold and titanium dioxide NPs to achieve a synergistic radiosensitization effect in vitro using triple-negative breast cancer SUM159 cells. 125 Such technology translated with a significant therapeutic effect both on tumor growth and animal survival in SUM159 tumor-bearing mice. 125  Clinicaltrials.gov was used to perform the search using the following terms: radiation, cancer, NPs (Table 2). Trials involving NPs used as a mean of drug delivery only were excluded.

| Trials involving AGuIX
The first phase I trial (NANORAD, NCT02820454) was dedicated to patients with multiple brain metastasis from non-small cell lung cancer (NSCLC), breast cancer, colon cancer, or melanoma and therapy consisted of whole brain radiotherapy (WBRT) (10 Â 3Gy/fraction over 3 weeks) combined with IV AGuIX nano-radioenhancers. 129  This initial success led to the NANORAD2 phase II trial (NCT03818386) that is currently recruiting patients and consists of 3 Â 100 mg/kg AGuIX IV injections (7 days prior to WBRT, before the 1st fraction and before the 6th fraction; 30 Gy and 3 Gy/fraction over 2-3 weeks). The primary endpoint compared to WBRT alone in this randomized trial is assessment of brain disease response at 3 and 6 months using RECIST (Response Evaluation Criteria in Solid Tumors).
A second single arm phase II trial (NANOSTEREO, NCT04094077) aims to assess the efficacy of AGuIX activated by fractionated stereotactic radiotherapy (SRT) in treatment of brain metastasis. Specifically, 100 mg/kg AGuIX is injected intravenously at days 4 and 8 followed by SRT on days 8 to 15 according to standard regimen. Similarly, the primary endpoint is brain metastases' response using RECIST. This trial has been terminated and a new trial design, including a second arm (placebo control group), is currently recruiting (NCT04899908).

| Trial involving Ferumoxytol (superparamagnetic iron oxide NPs)
In November 2020, a phase I prospective observation study has been launched for the use of Ferumoxytol, that is, iron oxide NPs, to enhance radiotherapy using a MR-Linac in the treatment of primary or metastatic hepatic cancers and liver cirrhosis (NCT04682847, phase I). The MR-Linac enables both the visualization of the tumor and NPs via MRI, as well as the delivery of radiation therapy.

| CONCLUSION AND FUTURE CHALLENGES
In this review, we discussed the recent advances involving the use of NPs as radiation therapy enhancers, both in preclinical and clinical studies. Such nanoscale technologies still face some critical biological barriers when injected intravenously such as nonspecific biodistribution, clearance by the reticuloendothelial system, hemorheological considerations, and cell internalization. 139 The use of active targeting strategies has shown promising success (vs. nontargeted nanoformulations) in the context of radioenhancement and drug delivery. However, the vast majority of the injected dose remains inefficiently delivered and is thus cleared. This led some investigators to use the intra-tumoral route for tumors of known location and, importantly, that are within reach. However, this translates into a limited number of eligible cancers. Thus, the next generation of nano-radioenhancers would benefit from improved biodistribution and tumor retention. A greater retention might enable a reduction in frequency of injections before each radiation fraction.
These efforts could potentially be achieved by combining strategies such as keeping the macrophages in clearance organs occupied prior to nanomedicine injection (such as demonstrated by Germain et al. 34 ), and leveraging the shape of NPs for improved targeting and/or retention (such as the strategies described in references, 25,73 respectively). In addition, approaches leveraging the intra-operative delivery of nanoformulations post tumor resection would be of particular interest to eliminate potential residual tumor/tumor margins using NPsensitized radiation. Such strategy has recently been developed for glioblastoma patients by Grauer et al. 140 An iron oxide NP paste was applied to the tumor resection site prior to combined hyperthermia and radiation therapies. 140

CONFLICT OF INTERESTS
Authors declare that they have no conflict of interest.

DATA AVAILABILITY STATEMENT
Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study.