FLASH radiotherapy: bridging revolutionary mechanisms and clinical frontiers in cancer treatment – a narrative review

Radiotherapy (RT) is a crucial component of antitumor therapies, and approximately 30% of cancer patients in Korea undergo RT [1]. RT must strike a balance between efficiently killing tumor cells and minimizing damage to normal tissue [2]. This constraint complicates the administration of an adequate tumoricidal dose, presenting a significant challenge, especially in the context of repeated RT [3,4]. Although numerous biological studies have been conducted to prevent or mitigate RT-induced acute and late toxicities, there have also been significant advancements in radiation technology over the past few decades, such as intensity-modulated RT, stereotactic body RT, and adaptive RT [5–10]. Moreover, technologies aided by artificial intelligence have been integrated into these advancements [11]. Despite these improvements, the radiation oncology community continues to explore new methods to expand the therapeutic window, as there remain unmet medical needs.

One of these innovative approaches is FLASH radiotherapy (FLASH-RT), which delivers irradiation at an ultra-high dose rate (UHDR) exceeding 40 Gy in less than a second [12]. This method can significantly shorten treatment times compared to conventional RT. In 1959, Dewey and Boag first observed that the radiosensitivity of Serratia marcescens decreased when exposed to 1.5 MV X-rays at a UHDR of 10–20 kilorads/2 µs, thereby protecting the bacteria compared to exposure at conventional dose rates—a phenomenon now known as the FLASH effect [13]. The term FLASH-RT was first introduced in a 2014 study by Favaudon et al. [14]. Subsequent preclinical experiments with mammalian cells and animal models have shown that UHDR irradiation, compared to conventional RT, provides a similar antitumor effect while also protecting normal tissue [12]. In 2019, a case report detailed the treatment of a patient with T-cell cutaneous lymphoma using FLASH-RT, noting that the approach was feasible and resulted in favorable outcomes for both the lymphoma and normal skin [15].

When cells and tissues are irradiated, a series of physical, chemical, and biological reactions occur. However, after FLASH-RT, these reactions do not advance to the biochemical phase [16]. Several radiobiological hypotheses have been proposed to explain the differential effects of FLASH-RT on normal and tumor tissues, including oxygen depletion, DNA damage, and the immune/inflammatory response [17]. Nevertheless, the exact mechanism of action of FLASH-RT is still not well understood.

The purpose of this review is to explore the revolutionary advancements and underlying mechanisms of FLASH-RT in cancer treatment. By integrating findings from both preclinical and clinical studies, this review aims to highlight the therapeutic potential and challenges of FLASH-RT, ultimately bridging the gap between cutting-edge radiobiological research and clinical application.

The initial studies on FLASH-RT were performed using low-energy electrons (~25 MeV) from either experimental or medical linear accelerators [20]. Conventional C-arm and intraoperative RT devices have also been successful in achieving UHDR [21]. These devices are readily available for UHDR, and several vendors are developing electron beam FLASH-dedicated linear accelerators [18]. However, the clinical application of this technology is restricted by the inherent properties of electrons, which include a low penetration depth of only a few centimeters, a short source-target distance (~50 cm), and a significant lateral penumbra [18,21]. Consequently, only skin cancers or tumors located within 2–3 cm of the body surface are currently suitable for treatment with FLASH-RT [22].

The use of very high energy electrons (VHEE, 50–250 MeV) has been suggested as an effective method for delivering therapeutic doses to tumors deep within the body using external electron beam FLASH-RT [22]. VHEE beams demonstrate relative insensitivity to body inhomogeneities compared to protons [23]. Although VHEE beams are associated with high entrance and exit doses, employing multi-directional beams can offer a benefit by sparing the skin [23,24]. However, due to the technical challenges associated with electron acceleration, VHEE research is limited to a few facilities, including the Photo Injector Test facility at Deutsches Elektronen-Synchrotron in Zeuthen and the European Organization for Nuclear Research Linear Electron Accelerator for Research facility [21].

Currently used linear accelerators in RT are unsuitable for photon beam-based FLASH-RT [18]. The primary reason is the high inefficiency in converting the electron beam to a photon beam, largely due to electron heat deposition [21,22]. This inefficiency necessitates the generation of a factor of 1,000 more electrons than current equipment can handle, presenting a significant challenge that must be overcome. Furthermore, technology that can accelerate this vast quantity of electrons and convert them into photons is also required [22].

Unlike high-energy beams, X-rays with energies <1 MeV can achieve FLASH conditions through a synchrotron [18]. The European Synchrotron Radiation Facility was the first to demonstrate that a UHDR synchrotron light source could reduce brain injury in mice following whole-brain irradiation [25]. Johns Hopkins University successfully achieved UHDR (40–240 Gy/s) using two 150 kVp fluoroscopy systems [26]. Development projects for new accelerators specifically designed for conventional, high-energy photon UHDR beams are currently in progress. Notable examples include the superconductive linac (6–8 MeV) from the Chengdu THz Free Electron Laser group and the Pluridirectional High-energy Agile Scanning Electronic RT platform from the Stanford Linear Accelerator Center [18,21].

Protons, unlike electrons and photons, possess a unique physical property known as the Bragg peak, where they deposit most of their energy at a specific depth just before stopping. This characteristic enables proton therapy to concentrate the dose on the tumor site while reducing exposure beyond the Bragg peak, thereby minimizing the risk of side effects [27]. Clinical isochronous cyclotrons utilized in proton therapy can deliver intensities exceeding 60 Gy/s at a fixed energy [18]. With proton energies surpassing 200 MeV, these cyclotrons are capable of treating deep-seated tumors [21]. As a result, protons are considered the most advanced technology for the clinical application of FLASH therapy [18].

Proton therapy employs pencil beam scanning to accurately target tumor volumes. However, a significant limitation of proton-based FLASH therapy stems from this method of beam delivery, as existing technology cannot accommodate the rapid energy modulation needed. In clinical settings, it is difficult to achieve FLASH conditions across the entire tumor due to the current speed of 3D volumetric scanning. Each energy change in the scanning process takes about one second, which is too slow to satisfy FLASH criteria. To achieve FLASH conditions throughout the entire tumor, the scanning speed must be increased by at least two orders of magnitude [18].

Researchers are actively working to overcome this challenge and bring this technology into routine clinical use. Hybrid active-passive systems featuring patient-specific 3D range modulators are currently being tested in clinical facilities and offer extremely rapid irradiation times (<1 s). Additionally, laser-driven accelerators are under development at Helmholtz-Zentrum Dresden-Rossendorf [18].

The depth-dose distribution of heavy ions exhibits a Bragg peak similar to that observed with protons; however, unlike protons, heavy ions also display a tail region [21]. Heavy ions offer additional advantages in RT, including a sharper lateral penumbra and higher relative biological effectiveness [18,21]. Despite these advantages, the global number of heavy ion centers remains limited. Consequently, research on FLASH-RT using heavy ions has been relatively sparse, partly due to the technical challenges associated with the synchrotron accelerators needed to achieve UHDR [21]. Currently, the Heidelberg Ion Beam Therapy Center has achieved dose rates exceeding 50 Gy/s using both helium and carbon ions. Meanwhile, Gunma University Hospital has reached dose rates up to 195 Gy/s with carbon ions, and the GSI Helmholtz Center for Heavy Ion Research has exceeded 100 Gy/s [18].

The first groundbreaking proof of concept study in 2014, using a lung fibrosis mouse model, demonstrated that 17 Gy FLASH-RT prevented lung fibrosis and radiation-induced acute apoptosis in blood vessels and bronchi. At 30 Gy, FLASH-RT was less fibrogenic than conventional RT, although it still caused some pulmonary fibrosis [14]. The morphology of lung tissue exposed to 20 Gy of either single or 10-pulse (fractionated) FLASH resembled that of non-irradiated tissue, with only minimal neutrophil infiltration [31]. This finding was consistent with observations from 30 Gy FLASH whole-thorax irradiation [32]. In contrast, the conventional RT group showed additional changes, including thickening of the alveolar septum and interstitial hemorrhage [31].

The study conducted by Fouillade et al. provides compelling evidence that FLASH irradiation offers protective effects on lung tissue compared to conventional RT [33]. The researchers observed that FLASH irradiation significantly reduced DNA damage in irradiated lung tissues and decreased the proliferation of lung progenitor cells following injury. Through single-cell RNA sequencing and histological analyses, it was revealed that FLASH reduced the activation of pro-inflammatory genes, diminished progenitor cell proliferation, and curtailed stem cell senescence, all factors contributing to lung fibrosis. Notably, lungs treated with FLASH showed a greater potential for tissue regeneration, exhibiting fewer signs of persistent DNA damage and senescence compared to those treated with conventional RT. However, these protective effects were absent in Terc^–/– mice, which have notably shortened telomeres and deficient telomerase activity, underscoring the critical role of telomere integrity and progenitor cell populations in harnessing the full benefits of FLASH.

In a study comparing thorax-irradiated mice, the FLASH group exhibited an 81% lower risk of mortality compared to the conventional group [32]. By the end of the follow-up period, survival rates were 100% in the control group, 90% in the FLASH group, and 50% in the conventional group.

Montay-Gruel et al. were the first to show that after 10 Gy whole-brain irradiation in mice, spatial memory was impaired at conventional dose rates, but remained unchanged at dose rates of ≥100 Gy/s [34]. This preservation of memory was attributed to the reduced impact of FLASH-RT on neurogenesis in the subventricular zone of the hippocampus compared to conventional dose rates. Montay-Gruel and colleagues also demonstrated this neuroprotective effect using synchrotron-generated X-rays for FLASH-RT [25]. Additionally, radiation-induced reactive astrogliosis was less severe following FLASH treatment than with conventional RT [25,35]. In later experiments, FLASH-RT was associated with long-term neurocognitive benefits over a 6-month period, reduced neuroinflammation, and preserved neuronal structure [36]. Not only was astrogliosis less pronounced in the FLASH group compared to conventional RT, but there was also no significant increase in the number of activated microglia in the hippocampus or brain cortex [36,37]. The dendritic area, branches, and length, which were significantly reduced after conventional RT compared to the control, were maintained in the FLASH-irradiated brain [36]. Similar results were observed in the number, density, and volume of dendritic spines.

Although the juvenile mouse brain is known to be radiosensitive, the FLASH effect was still observed [38,39]. Further research has shown that FLASH-RT not only spares the neurogenic niche but also preserves pituitary function, as evidenced by stable growth hormone levels [38]. Even with hypofractionated regimens, it prevented radiation-induced neurocognitive complications in the normal brain [39,40]. Other researchers have also validated the relationship between FLASH-RT and reduced cognitive deficits in both spatial and non-spatial object recognition, as well as associated neurodegeneration, at the electrophysiological, molecular, and structural levels [39,41,42].

In experiments involving proton-based FLASH, no significant differences were observed in locomotion, exploratory behavior, spontaneous activity, or anxiety levels among the control, FLASH, and conventional RT groups [43]. However, in the object recognition task used to assess memory, both the control and FLASH groups demonstrated good recall of the familiar object, in contrast to the rats that received conventional RT. Analysis of the immune cell populations in the brain parenchyma revealed that infiltration of peripheral immune cells (CD45^high) occurred irrespective of the dose rate. Additionally, there was a 4-fold reduction in microglia compared to non-irradiated tissue, with no variations noted across different dose rates.

These beneficial effects of FLASH might result from reduced production of reactive oxygen species or less pronounced increases in pro-inflammatory cytokines following FLASH-RT [36,41]. They may be associated with the preservation of cerebrovascular integrity through the protection of tight junctions or aquaporin-4 levels [37,39,44].

FLASH-RT has the potential to mitigate radiation-induced skin reactions, which could significantly reduce both acute and late skin reactions in the treatment of head and neck cancers and extremity soft tissue sarcoma when integrated into future clinical practice [45–48]. When murine skin was subjected to UHDR proton therapy, the skin reaction score was lower compared to that observed with conventional proton therapy [45–47]. There were fewer instances of epidermal necrosis, skin stem cell depletion, hair follicle atrophy, inflammation, epidermal hyperplasia, myofiber atrophy, and bone remodeling [46,47,49,50]. The potential benefits of FLASH-RT have also been demonstrated in mini-pigs; however, severe late skin necrosis, which was volume-dependent, developed but eventually resolved [51,52]. Regarding lymphedema, no differences were noted in the incidence or progression between the dose rates; however, the severity was greater in mice that received conventional RT [46]. This reduction in side effects translated into a survival benefit for mice that received ≤40 Gy, although no survival difference was observed between the two dose rates at the 45 Gy dose [45,46].

Zhang et al. reported the oxygen dependence of the FLASH effect [48]. FLASH irradiation improved skin contraction (25–30 Gy), epidermal thickness (25 Gy), and collagen deposition (25 Gy). However, when irradiation occurred in a 100% oxygen environment or under hypoxic conditions induced by restricting blood flow through leg constriction, the tissue-sparing effect of FLASH was lost.

In transcriptome analysis, pathways such as apoptotic signaling, keratinocyte differentiation, and cornification were upregulated in the group that received conventional proton therapy. Conversely, these changes were not observed in the UHDR group [46]. TGF-β1 expression was also observed at lower levels following FLASH-RT [46,47]. The levels of chemokine ligand-1 and granulocyte-colony stimulating factor increased, while those of granulocyte-macrophage colony-stimulating factor decreased following conventional RT [47]. The granulocyte-macrophage colony-stimulating factor/granulocyte-colony stimulating factor ratio, which inversely correlates with tissue toxicity, was reduced in the conventional group, suggesting increased tissue toxicity [53]. In the FLASH group, IL-6 levels rose, although no significant differences in IL-6 levels were noted between the 57 Gy/s and 115 Gy/s dose rates [47]. However, when proton irradiation was delivered at 930 Gy/s, no differences were observed in the levels of TGF-β1, IL-1α, IL-1β, and tumor necrosis factor-α in the blood [50].

FLASH experiments using electrons demonstrated more favorable crypt survival at doses between 7.5–12.5 Gy after whole-abdominal irradiation than after conventional dose rate irradiation [54]. However, this sparing effect decreased as the number of FLASH pulses increased or as the interval between pulses extended, leading to a longer delivery time. When 15 Gy of proton irradiation was administered to the whole abdomen, there was a smaller reduction in proliferating cells within the jejunum crypts in the FLASH group than in the conventional RT group [55,56]. After intestinal irradiation at 15 or 18 Gy, the extent of intestinal fibrosis was similar to that observed in non-irradiated tissue [32,55].

Some studies have shown that abdominal FLASH irradiation reduces mortality in mice suffering from radiation-induced gastrointestinal syndrome compared to conventional irradiation [32,57]. This protective effect is believed to stem from FLASH irradiation's ability to decrease chromosomal damage and apoptosis in the crypt base columnar cells of the jejunum, thereby helping to preserve intestinal function and epithelial integrity [57]. Additionally, the beneficial impact of FLASH X-rays may be associated with differing inflammatory responses, including reduced activation of the cyclic guanosine monophosphate–adenosine monophosphate synthase-stimulator of interferon genes (cGAS-STING) pathway and changes in the redox status within the intestinal crypts [58,59].

Fecal samples were utilized for gut microbiome analysis, revealing that overall α-diversity and evenness declined across all irradiated groups, although richness decreased solely in the conventional group [54]. In the β-diversity analysis, the cluster of the FLASH-treated group was closer to that of the control group, suggesting fewer alterations in the microbiome.

The impact of FLASH-RT on the heart remains largely unexplored. Until recently, research in this area continued to be scarce, with the heart being an uncharted area in FLASH-RT studies. It was not until 2024 that the first study addressing the effects of FLASH-RT on cardiac tissue was published, marking a significant advancement in our understanding of how this innovative RT might influence cardiac function.

A recent study investigated the impact of proton FLASH-RT aimed specifically at the cardiac apex, delivering a precise 40 Gy dose [60]. The research utilized γH2AX staining to evaluate DNA damage, which was found to be limited to the lower third of the heart, with no impact on adjacent tissues. Bulk RNA sequencing of cardiac tissue revealed distinct pathway regulations based on the treatment approach. In the FLASH-RT group, pathways related to cytoplasmic translation, mitochondrion organization, and adenosine triphosphate synthesis were upregulated. In contrast, pathways involved in tissue morphogenesis and the regulation of developmental growth were downregulated. A key finding was that FLASH-RT reduced cardiac inflammation and profibrotic responses, leading to decreased myocardial fibrosis. Unlike conventional RT, FLASH-RT maintained heart functionality at levels similar to those of non-irradiated controls.

Tumor cell killing was not altered, as reported by Favaudon et al., who found that both xenograft human tumor and syngeneic orthotopic lung tumor models exhibited equivalent tumor growth inhibition when comparing FLASH-RT (4.5 MeV electrons, 60 Gy/s) with conventional RT [14]. Similarly, other studies employing various tumor models and FLASH sources have demonstrated comparable levels of histological tumor cell damage, regardless of the dose rate and fractionation [31,40,43,46,47,49,55–57,61–67]. In some instances, tumor growth was even more delayed with FLASH-RT than with conventional RT [32]. In a separate study using 250 MeV proton beams, no difference in lung tumor diameter was observed between 18 Gy FLASH (60 Gy/s) and conventional irradiation; however, there was a significant reduction in proliferating tumor cells following FLASH, indicating a meaningful decrease in lung tumor burden [68]. The survival of tumor-bearing mice was found to be equivalent to or better with FLASH-RT compared to conventional RT [14,32,40,56,66].

Although numerous studies have explored the interaction between the tumor immune microenvironment (TIME) and FLASH-RT, the results have so far been varied and inconsistent [43, 61, 62, 65, 68]. Some researchers have observed that FLASH retains its antitumor efficacy even in severely immunodeficient mice, suggesting the existence of an antitumor mechanism that may function independently of the immune response [65]. FLASH-RT has been shown to enhance cytotoxic T-cell infiltration into tumors and reverse the immunosuppressive phenotype [68]. There was an increase in CD8⁺ T-cell recruitment to the tumor, accompanied by a decrease in the infiltration of immunosuppressive regulatory T-cells (Treg). Additionally, macrophage polarization shifted towards an M1-like phenotype, which facilitated increased lymphocyte infiltration in lung tumors. Furthermore, FLASH-RT suppressed the expression of programmed death-1 (PD-1) and its ligand (PD-L1).

In an orthotopic glioma rat model, tumor-infiltrating lymphocytes, including CD4⁺ and CD8⁺ T-cells, increased at both conventional dose rates and UHDR (226 MeV proton, 257 Gy/s) [43]. Interestingly, Treg levels also increased in both groups. Additionally, there were observed increases in natural killer cells and B cells, suggesting that cranial irradiation activates adaptive immunity. However, in the FLASH group, no increase in tumor myeloid cells was noted.

A very recent study using an orthotopic syngeneic mouse model of brainstem diffuse midline glioma explored high-resolution profiling of the TIME following FLASH (9 MeV electron, 90 Gy/s) and conventional dose-rate RT [61]. The methods employed included single-cell RNA sequencing and flow cytometry. Analysis of CD45⁺ cells revealed that both the FLASH and conventional groups displayed similar proportions of immune subsets, with microglia as the predominant population. As an acute effect of RT, both FLASH and conventional irradiation triggered a type 1 interferon (IFN1) response in microglia. However, by day 10 post-RT, the FLASH group exhibited a dose-rate-dependent reduction in the IFN1 response in microglia, indicating a distinct temporal pattern and suggesting that microglial activation by FLASH was transient during the early stages. Regarding non-resident myeloid cells, such as macrophages and dendritic cells, which represented a minor fraction of the TIME, an early IFN1 response was observed in the conventional group, but in the FLASH group, this response was not clearly defined until day 10 post-RT. Despite these temporal immune changes, no significant differences in tumor control were noted between the dose rates, highlighting an area for future research.

Several trials have explored FLASH-RT in animals with cancer, using electrons in the 4.5-12 MeV range [51,52,69–71]. In one study, seven cats with T1/2N0M0 squamous cell carcinoma of the nasal planum received 30 Gy of radiation [52]. All remained tumor-free for one year, with only one case showing progression thereafter. Another trial involved six cats with locally advanced T2/T3N0M0 tumors treated with 25–41 Gy, achieving an 84% progression-free survival rate at 16 months [51]. Additionally, a collaborative effort between researchers from Denmark and Sweden applied FLASH-RT to canine cancer patients with superficial malignant tumors [69–71]. The treatment was effective, although it was associated with a potential risk of osteoradionecrosis.

The oxygen depletion hypothesis is currently the most widely accepted theory. It is based on the principle that oxygen acts as a critical radiosensitizer in RT; thus, tissues with a high oxygen supply are more radiosensitive [75]. FLASH irradiation rapidly depletes oxygen, leaving insufficient time for oxygen to be replenished from the surrounding circulating blood [74]. This results in acute hypoxic conditions that lead to transient radioresistance, thereby sparing normal tissue [76]. Conversely, tumors, with their inherently abnormal blood vessels, are already adapted to hypoxic conditions. This adaptation explains why the dose rate does not significantly impact the tumor cells' susceptibility to radiation [17].

A limitation of this hypothesis is that while FLASH-RT resulted in greater oxygen consumption compared to conventional RT, it did not completely deplete all the oxygen [77]. Furthermore, the oxygen levels associated with higher cell survival rates following FLASH-RT vary significantly across experiments, ranging from severely hypoxic conditions (<0.5%) to oxygen-rich environments like those found in the lungs [74]. This variability suggests that the oxygen depletion hypothesis may not fully explain the FLASH effect.

An alternative explanation has been proposed, suggesting that reactive oxygen species, which serve dual roles as signaling and damaging agents within cells, may interact with molecules involved in redox metabolism. This interaction could potentially play a pivotal role in the FLASH effect [18].

Cell fate after irradiation is primarily determined by DNA damage, specifically unrepaired DNA double-strand breaks [78]. Several studies have shown that DNA damage is less severe after FLASH irradiation [33,37,57]. This reduction in DNA damage helps preserve stem and progenitor cells across various tissues, consequently decreasing toxicity [33,34,38,46,55,56]. However, while this effect accounts for the sparing of normal tissue, it does not completely explain the sustained antitumor activity. Although the precise mechanisms are still not fully understood, it is possible that differences in the activation of downstream pathways after DNA damage—such as DNA repair pathways, the cGAS-STING pathway, or the immune system—between normal and tumor cells could contribute [17].

FLASH significantly reduces the duration of radiation exposure, which is anticipated to decrease the volume of irradiated blood and aid in preserving circulating immune cells from depletion [50,79]. However, several in vivo studies have yielded negative results, showing no significant difference in the circulating immune cell populations between FLASH and conventional dose rates [43,59,80]. Instead, while further detailed research is necessary, it is generally observed that there is an increase in cytotoxic T-cell infiltration into tumors [43,68]. Conversely, the persistence of the antitumor effect in immunocompromised animals suggests that this effect cannot be solely attributed to the immune response [65].

FLASH also reduces TGF-β and pro-inflammatory gene expression, as well as the release of pro-inflammatory cytokines, thereby mitigating stress response and inflammation [33,41,46,47,58–60,81]. This contributes to the preservation of normal tissue, exemplified by the reduction of neuroinflammation in the normal brain, which in turn supports the maintenance of neurological function [25,35–37]. Given that the TGF-β pathway is a pharmacological target in cancer therapy, the reduction of TGF-β expression induced by FLASH could enhance antitumor activity, similar to the effects of TGF-β antagonists [82].

Current cancer treatment is witnessing a revival of interest in immunotherapy, particularly in its integration with RT [83]. In this context, the immune response elicited by FLASH provides compelling insights that may herald a new phase in radioimmunotherapy. First, FLASH reduces the expression of PD-1 and PD-L1, thereby inhibiting the tumor's ability to evade the immune system [68,84]. In a study using an ovarian cancer mouse model, abdominopelvic irradiation followed by PD-1 therapy led to enhanced tumor control in both conventional and UHDR settings, without an increase in toxicity compared to using FLASH alone [85]. This treatment also resulted in a lower Treg-to-T-effector ratio and a higher level of CD8⁺ T-cell infiltration within the tumor. While immunotherapy alone often yields only modest response rates, these findings are noteworthy as they indicate that FLASH-RT can significantly improve the effectiveness of PD-1/PD-L1 inhibitors.