Preclinical validation of a clinical prototype of intraoperative electron FLASH radiotherapy equipment: key evidence for a new radiotherapy paradigm Authors: Xiaofei Hao, Huan Du, Binwei Lin, Yang Xia, Mingming Tang, Wei Wu, Decai Wang, Yihan Zhu, Yu Zhang, Jianxin Wang, Yiwei Yang, Dai Wu, Xiaobo Du Published: 2025 ABSTRACT Background and purposeThis study aimed to evaluate whether the clinical prototype of intraoperative electron FLASH radiotherapy (CPIO-EFLASH), with a source-surface distance (SSD) of 50 cm, could achieve ultra-high dose rates, effectively control tumors, and trigger the FLASH tissue-sparing effect in preclinical models.Materials and methodsSubcutaneous tumor-bearing mice (4T1 breast cancer, U87-MG glioma, PAN02 pancreatic cancer) and healthy C57BL/6 female mice (whole-brain, thorax, abdomen and single-leg irradiation) were subjected with ultra-high dose rate radiotherapy (UHDR-RT, ≥ 40 Gy/s), conventional doserate radiotherapy (CONV-RT,0.07 Gy/s), or sham radiotherapy (Control). We evaluated survival status, tumor growth suppression, apoptosis, proliferation, and DNA damage in tumor tissues, along with radiation-induced injuries to the brain, lung, intestine, and skin tissues.ResultsThe actual dose rates of UHDR-RT ranged from 192 to 473 Gy/s. No significant difference in tumor growth suppression was observed between the UHDR-RT and CONV-RT. Two months post whole-brain irradiation, UHDR-RT demonstrated better spatial learning and memory abilities compared to CONV-RT. At 120 days post whole-thorax irradiation and 90 days post whole-abdomen irradiation, the survival rates of UHDR-RT were also significantly higher. Histological analyses revealed more severe injury to lung and intestinal tissues in the CONV-RT group. Additionally, UHDR-RT exhibited milder radiation-induced skin injury from 2 to 8 weeks post-irradiation.ConclusionThe CPIO-EFLASH can achieve ultra-high dose rates (≥ 40 Gy/s at an SSD of 50 cm) and trigger significant normal tissue-sparing effects. Integrating electronic FLASH technology into intraoperative radiotherapy may bring potential clinical benefits by effectively treating tumors, while minimizing radiation-induced injury to normal tissues. Our findings highlight the necessity for further clinical trials of CPIO-EFLASH in intraoperative radiotherapy.Supplementary InformationThe online version contains supplementary material available at 10.1186/s13014-025-02781-x. Introduction The incidence of cancer is increasing annually, becoming one of the leading causes of human mortality [1]. Radiotherapy has been one of the most widely used treatments for both early-stage and advanced cancerpatients. It primarily destroys the genetic material of cancer cells through various types of radiation, either alone or in combination with surgery or chemotherapy for treatment. However, while targeting tumors, radiation can also damage surrounding normal tissues, leading to adverse effects such as radiation-induced brain injury (estimated incidence,50%–90%) [2], lung injury (40%) [3], intestinal injury (50%) [4], and skin injury (85%) [5], which can severely impact patients’ quality of life. In recent decades, radiation toxicity has emerged as a global concern, with minimizing adverse effects while maintaining therapeutic efficacy as a key objective in clinical radiotherapy. Intraoperative radiation therapy (IORT) is a cancer treatment method with over 50 years of clinical application [6]. It combines radiotherapy with surgical treatment, delivering a single high dose of radiation directly to the tumor bed, visible tumor, or areas prone to recurrence and metastasis. The primary advantages of IORT include improved local tumor control rates, reduced damage to healthy tissues through precise irradiation, and shortened treatment cycles, thereby eliminating the need for multiple visits associated with traditional radiotherapy. Additionally, advancements in mobile radiotherapy equipment and imaging technology have enhanced the safety and precision of IORT, gradually establishing it as an important treatment modality for various cancers, including gliomas, breast cancer, and pancreatic cancer [7]. Despite its precision, IORT may still damage surrounding normal tissues and cause complications. For instance, IORT for breast cancer can also lead to the occurrence of skin toxicity [8, 9]. Moreover, as it involves a combination of surgery and radiotherapy, the extended duration of in-operating room radiation time also increases the patient’s anesthesia duration and elevates surgical risks. Ultra-high dose rate radiotherapy (UHDR-RT), known as FLASH radiotherapy, is an emerging beam radiotherapy technique delivering doses at an extremely high rate (≥ 40 Gy/s). Compared to conventional dose rate radiotherapy (CONV-RT, ≤ 0.1 Gy/s), UHDR-RT’s primary advantage lies in its ability to reduce damage to normal tissues surrounding the tumor while maintaining equivalent antitumor efficacy, a phenomenon termed the FLASH effect [10, 11]. By expanding the therapeutic window for tumors, UHDR-RT enables higher dose delivery to cancerous tissues while better protecting normal tissues, thereby enhancing antitumor efficacy. Additionally, UHDR-RT can complete the entire treatment process within a very short time, significantly reducing the treatment time consumption compared to CONV-RT. The radiation sources currently utilized in UHDR-RT include electrons [12, 13], protons [14, 15], and photons [16, 17]. High-energy photons are the predominant radiation source in clinical radiotherapy, however, achieving UHDR photon beams remains technically challenging. Proton beam radiotherapy is well-suited for treating deep-seated tumors. Electron beams, which are cost-effective and less technically demanding, have limited penetration depth and are primarily used for superficial tumors [18] and intraoperative radiotherapy [19]. Moreover, electrons are the most extensively researched UHDR-RT radiation source worldwide. Numerous preclinical studies have demonstrated that electron UHDR-RT achieves protective effect on normal tissues, including the lungs [11], intestines [20], skin [21], and brain [12]. In 2019, Swiss researchers conducted the first clinical study of electron UHDR-RT, in which patients with cutaneous T-cell lymphoma were treated at a dose rate of 166.7 Gy/s. The irradiated tumors achieved complete remission, with only grade 1 skin toxicity observed [22]. Therefore, applying electron UHDR-RT technology to IORT could significantly reduce damage to normal tissues and radiation-related complications, as well as shorten the treatment time in clinical patients. Consequently, our team have developed a clinical prototype of intraoperative electron FLASH radiotherapy (CPIO-EFLASH, Zhongjiu Flash Medical Technology Co., Ltd.) suitable for clinical use, with a source-surface distance (SSD) of up to 50 cm. The device integrates the radiation source with a six-degree-of-freedom movable robotic arm, and it will be user-friendly and portable, offering user-friendly and portable features to reduce operational complexity for clinicians. This study aimed to evaluate whether CPIO-EFLASH can effectively treat preclinical tumor models of glioblastoma, breast cancer, and pancreatic cancer while providing protective effects for the brain, lungs, intestines, and skin. The findings will provide critical evidence to support subsequent clinical trials. Materials and methods Irradiation setup The UHDR-RT experiments were conducted at the CPIO-EFLASH facility in Mianyang, China. The facility weighs 2680 kg, has a minimum height of 3.2 m, and covers an area of at least 36 m². The facility’s superconducting linear accelerator with SSD of 50 cm, can generate electrons with energies ranging from 9 MeV to 12 MeV. In this study, 9 MeV was consistently used for all ultra-high dose rate radiotherapy (UHDR-RT), and this energy level fulfills the requirements of all irradiation targets in the research. Stray X-rays account for no more than 2% of the total dose. Lead collimators with apertures of 1 × 1.5 cm² (for whole-brain irradiation,10 Gy*1F),3 × 2 cm² (for whole-thorax irradiation,30 Gy*1F),4 × 4 cm² (for whole-abdomen irradiation,13 Gy*1F), and 2 × 2 cm² (for tumor irradiation,10 Gy/16 Gy*1F and single-leg skin irradiation,24 Gy*1F) were used to define the FLASH irradiation area (Fig. 1B). Figure 1A illustrates a schematic diagram of the irradiation device used for irradiating mice. Experimental mice were anesthetized using isoflurane and immobilized on a custom-made clear polymethyl methacrylate (PMMA) with a thickness of 0.5 cm (Fig. 1C, Supplementary Fig. 2A). The laser positioning system with vertical and horizontal lasers were employed to align the beam center, ensuring that the irradiated tissue or tumor was precisely positioned at the center of the radiation field. The CONV-RT experiments were conducted using the clinical 6 MeV Elekta Infinity linac system (Elekta AB, Stockholm, Sweden). Given the uniform and stable dose output of CONV-RT, we employed radiation fields of 24 cm × 1.5 cm (brain), 24 cm × 2 cm (thorax, skin, and tumor), and 24 cm × 4 cm (abdomen) for CONV-RT to target different regions respectively. Fig. 1Iradiation setup, parameters, and dosimetry characteristics. (A) Schematic diagram of the irradiation setup for electron ultra-high dose rate radiotherapy (UHDR-RT). (B) Field sizes of the UHDR-RT collimator for different irradiation sites in mice: whole-brain irradiation, 1 × 1.5 cm²; whole-thorax irradiation, 3 × 2 cm²; whole-abdomen irradiation, 4 × 4 cm²; skin and tumor irradiation, 2 × 2 cm². (C) Example of irradiation setup for a U87-MG glioma tumor model mouse. A 5-mm-thick polymethyl methacrylate (PMMA) plate was used for mouse fixation. EBT-XD film wais placed between the PMMA plate and the mouse’s surface to evaluate the dose. The laser positioning system with vertical and horizontal lasers aligns the mouse with the beam center. (D) EBT-XD film 24 h post-irradiation (U87-MG,4T1,10 Gy). (E) Representative relative spatial dose distribution of UHDR-RT measured with EBT-XD film (U87-MG,4T1,10 Gy). (F) Representative relative spatial dose distribution of CONV-RT measured with EBT-XD film (U87-MG,4T1,10 Gy). (G) Horizontal and vertical dose curves of the dose distribution of UHDR-RT and CONV-RT (U87-MG,4T1,10 Gy; The position of mice was marked with a green frame; sufficient lateral clearance was maintained in the UHDR-RT device to ensure dose uniformity between two groups.). (H) The PDD (Percent Depth Dose) curve of electrons sources for UHDR-RT and CONV-RT (U87-MG,4T1,10 Gy), with the mouse (green frame) placed 5 mm from the collimator with a 5-mm build up area depth. A 10 Gy administered 2.5 mm under the skin, with comparable PDD curves between the FLASH-RT and CONV-RT groups Iradiation setup, parameters, and dosimetry characteristics. (A) Schematic diagram of the irradiation setup for electron ultra-high dose rate radiotherapy (UHDR-RT). (B) Field sizes of the UHDR-RT collimator for different irradiation sites in mice: whole-brain irradiation, 1 × 1.5 cm²; whole-thorax irradiation, 3 × 2 cm²; whole-abdomen irradiation, 4 × 4 cm²; skin and tumor irradiation, 2 × 2 cm². (C) Example of irradiation setup for a U87-MG glioma tumor model mouse. A 5-mm-thick polymethyl methacrylate (PMMA) plate was used for mouse fixation. EBT-XD film wais placed between the PMMA plate and the mouse’s surface to evaluate the dose. The laser positioning system with vertical and horizontal lasers aligns the mouse with the beam center. (D) EBT-XD film 24 h post-irradiation (U87-MG,4T1,10 Gy). (E) Representative relative spatial dose distribution of UHDR-RT measured with EBT-XD film (U87-MG,4T1,10 Gy). (F) Representative relative spatial dose distribution of CONV-RT measured with EBT-XD film (U87-MG,4T1,10 Gy). (G) Horizontal and vertical dose curves of the dose distribution of UHDR-RT and CONV-RT (U87-MG,4T1,10 Gy; The position of mice was marked with a green frame; sufficient lateral clearance was maintained in the UHDR-RT device to ensure dose uniformity between two groups.). (H) The PDD (Percent Depth Dose) curve of electrons sources for UHDR-RT and CONV-RT (U87-MG,4T1,10 Gy), with the mouse (green frame) placed 5 mm from the collimator with a 5-mm build up area depth. A 10 Gy administered 2.5 mm under the skin, with comparable PDD curves between the FLASH-RT and CONV-RT groups Dosimetry monitoring The beamline current was monitored using a current sensor, while the electron beam dose detector was installed in the ionization chamber upstream of the collimator. During irradiation, Gafchromic™ EBT-XD radiochromic films (Ashland Inc., Covington, Kentucky, USA) (Fig. 1D, Supplementary Fig. 1A, D, G, Supplementary Fig. 2B) were used as passive dosimeters to measure the absolute dose and beam profile. These films, which fully covered the irradiation field, were placed on the anterior surface of the mice (Fig. 1B and C). The RGB values of the films were scanned 5 min post-irradiation using an Epson Expression 12000XL scanner (Seiko Epson Corporation, Japan), and analyzed using SNC patient software (Sun Nuclear Corporation, USA). Dose of the horizontal and vertical axes were generated to evaluate whether the dose magnitude and distribution are uniform. The percent depth dose (PDD) curve clarifies the dose distribution of the radiation beam within the depth range of the target volume, ensuring that the target volume receives the preset prescribed dose. Additionally, before the UHDR-RT experiment, the EBT-XD film was also calibrated using a clinically applied 6 MeV Elekta Precise linac. This dose monitoring method was similar to previous reports [16]. Mice and tumor model Six-week-old female mice were obtained from GemPharmatech Co., Ltd. (Jiangsu, China) and SPF (Beijing) Biotechnology Co., Ltd. (Beijing, China) and housed in the Specific Pathogen Free (SPF) animal laboratory at Mianyang Central Hospital (Mianyang, China). All experiments were conducted following a 1-week acclimatization period. Throughout the study, mouse welfare was strictly maintained in compliance with relevant ethical guidelines, and the experimental protocol was approved by the Animal Ethics Committee of Mianyang Central Hospital (approval number: P2020032). The detailed methods are provided in the supplementary materials. Histological analyses Acute radiation-induced lung injury was assessed based on the number of alveoli and alveolar area [23], and lung injury pathological score [24]. Acute radiation-induced intestinal injury was quantified based on the number of intestinal crypts per unit area [25], villus height, and intestinal injury pathological score [25]. Due to insufficient sample size at the survival endpoint (only one mouse remaining in the CONV-RT group), only Masson’s trichrome staining was performed without statistical analysis. Skin injury was assessed based on the prevalence of skin moist desquamation [26], epidermal thickness, dermal thickness [17], skin injury score [26], and relative collagen per unit area (100%) [17]. Statistical analyses Statistical analyses were conducted using Prism version 8.4.0 (GraphPad Software). For comparisons among three groups, one-way analysis of variance (ANOVA) was performed, followed by Tukey’s multiple comparisons test. For comparisons between two groups, Student’s t-test was used. Kaplan–Meier curves were constructed to summarize survival probabilities, and differences were assessed using the log-rank test. All data are expressed as the mean ± standard deviation (SD). Statistical significance was set at P < 0.05. Results Dosimetry The experimental parameters for the irradiation of normal tissues and tumors in this study are detailed in Supplementary Table 1. The average dose rates for this study ranged from 192 to 473 Gy/s. The actual irradiation doses for the two irradiation groups are as follows: UHDR-RT group (10 Gy/brain, 30.73 ± 0.74 Gy/thorax, 13.29 Gy/abdomen, 23.36 ± 0.32 Gy/skin, 10.25 ± 0.07 Gy/4T1, 10.19 ± 0.03 Gy/U87-MG, 15.99 ± 0.08 Gy/PAN02); CONV-RT group (10 Gy/brain, 30 Gy/thorax, 13.29 Gy/abdomen, 23.5 Gy/skin, 10.30 Gy/4T1, 10.20 Gy/U87-MG, 16 Gy/PAN02). The dose difference between these two groups was less than 5%. The relative spatial dose distributions of UHDR-RT and CONV-RT were measured using EBT-XD film. When irradiating different tumors and tissues, the horizontal and vertical dose curves, as well as the PDD curves of the dose distribution for these two groups essentially coincide (Fig. 1E and H, Supplementary Fig. 1, Supplementary Fig. 2). UHDR-RT reduced radiation-induced brain injury Two months after whole-brain irradiation, three groups of mice underwent the Morris Water Maze test (Fig. 2A). The CONV-RT group (8.5 ± 1.91 s) spent less time in the target quadrant compared to the UHDR-RT group (13.2 ± 3.27 s) and the Control group (16.9 ± 2.23 s), indicating impaired spatial memory accuracy in the CONV-RT group (P < 0.05) (Fig. 2B). Figure 2C shows the escape latency trends over the four-day period, reflecting the learning process of the animals. The CONV-RT group exhibited an increase in escape latency on the second day, while both the UHDR-RT and Control groups showed consistent decreases each day. The Control group maintained the shortest escape latency throughout the test. Notably, on the second day, the UHDR-RT group (54.85 ± 1.56 s) had a shorter escape latency than the CONV-RT group (58.95 ± 1.64 s), suggesting superior learning ability in the UHDR-RT group (P < 0.05). However, no significant differences were observed between the UHDR-RT and CONV-RT groups on the other three days. Supplementary Fig. 3 showed the mouse’s swimming trajectory of the Morris Water Maze test. Fig. 2Brain injury post whole-brain irradiation. (A) Schematic diagram of whole-brain irradiation and the Morris Water Maze test. (B) Duration target quadrant images for three groups of mice followed up for 60 days after whole-brain irradiation (UHDR-RT: 10 Gy, 250 Gy/s, n = 6; CONV-RT: 10 Gy, 0.07 Gy/s, n = 6; Control: sham radiotherapy, n = 6). (C) Comparison of Escape latency in the Morris Water Maze test for three Groups over four consecutive days (line chart on the left, bar chart on the right). *p < 0.05, **p < 0.01, ***p < 0.001 Brain injury post whole-brain irradiation. (A) Schematic diagram of whole-brain irradiation and the Morris Water Maze test. (B) Duration target quadrant images for three groups of mice followed up for 60 days after whole-brain irradiation (UHDR-RT: 10 Gy, 250 Gy/s, n = 6; CONV-RT: 10 Gy, 0.07 Gy/s, n = 6; Control: sham radiotherapy, n = 6). (C) Comparison of Escape latency in the Morris Water Maze test for three Groups over four consecutive days (line chart on the left, bar chart on the right). *p < 0.05, **p < 0.01, ***p < 0.001 UHDR-RT improved survival probability and reduced radiation-induced lung injury The schematic diagram of whole-thorax irradiation is presented in Fig. 3A. 120 days following whole-thorax irradiation, the survival rate of the CONV-RT group (10%) was significantly lower than that of the UHDR-RT group (80%) and the Control group (100%) (P < 0.001; Fig. 3B). In the lung tissue hematoxylin and eosin (H&E) staining pathological sections obtained 72 h post-irradiation, the CONV-RT group exhibited disordered lung tissue structure, hemorrhage, and inflammatory cell infiltration (Fig. 3C), along with a significant reduction in the number and area of alveoli (Fig. 3D). The lung tissue damage scores for the three groups were as follows: Control group,6.67 ± 14.14; UHDR-RT group, 36; and CONV-RT group,65.33 ± 14.14, with statistically significant differences observed (P < 0.001). Given that only one mouse in the CONV-RT group survived beyond 120 days, Masson staining was performed on the lung tissue without statistical analysis. The sections revealed that the degree of lung fibrosis was less severe in the UHDR-RT group compared to the CONV-RT group. UHDR-RT improved survival probability and reduced radiation-induced intestinal injury Figure 3E showed the schematic diagram of whole-abdomen irradiation.90 days post irradiation, the survival rate of the CONV-RT group (10%) was significantly lower than that of the UHDR-RT group (80%) and the Control group (100%) (P < 0.0001; Fig. 3F), with the peak mortality period occurring 5–7 days post-irradiation. Seventy-two hours after irradiation, small intestine tissues were harvested for H&E staining. The pathological sections revealed that the CONV-RT group exhibited disordered intestinal structure (Fig. 3G), with the fewest crypts per millimeter and the lowest villus height (Fig. 3H). The intestinal tissue damage scores were as follows: Control group0; UHDR-RT group,1.2 ± 0.45; CONV-RT group,2.8 ± 0.45, with statistically significant differences observed (P < 0.05). Similar to the whole-thorax irradiation group, only one mouse in the CONV-RT group survived beyond 90 days following whole-abdomen irradiation. Consequently, Masson staining was performed on the small intestine tissues without statistical analysis. The sections demonstrated no significant intestinal fibrosis in any of the three groups. Fig. 3Survival percentage, lung and intestinal injury post whole-thorax and whole-abdomen irradiation. (A) Schematic diagram of whole-thorax irradiation and observation indicators in mice. (B) Kaplan–Meier survival curves for three groups of mice followed up for 120 days after whole-thorax irradiation (UHDR-RT,30 Gy,473 Gy/s, n = 10; CONV-RT,30 Gy,0.07 Gy/s, n = 10; Control, n = 10). (C) Representative images of histopathological changes in lung samples (H&E staining, 72 h post-irradiation, scale bars: 100×, 200 μm, 400×, 50 μm; Masson staining, 120 days post-irradiation, scale bars: 100×, 200 μm, 400×, 50 μm). (D) Quantitative analysis of the number and area of alveoli and lung injury scores in H&E staining sections from the three groups. (E) Schematic diagram of whole-abdomen irradiation and observation indicators in mice. (F) Kaplan–Meier survival curves for three groups of mice followed up for 90 days after whole-abdomen irradiation (UHDR-RT,13 Gy,300 Gy/s, n = 10; CONV-RT,13 Gy,0.07 Gy/s, n = 10; Control, n = 10). (G) Representative images of histopathological changes in intestinal samples (H&E staining, 72 h post-irradiation, scale bars: 12.5×, 250 μm, 200×, 100 μm; Masson staining, 90 days post-irradiation, scale bars: 12.5×, 250 μm, 200×, 100 μm). (H) Quantitative analysis of the number of crypts per millimeter, villus height, and intestinal injury scores in H&E staining sections from the three groups. *p < 0.05, ** p < 0.01, ***p < 0.001, ****p < 0.0001 Survival percentage, lung and intestinal injury post whole-thorax and whole-abdomen irradiation. (A) Schematic diagram of whole-thorax irradiation and observation indicators in mice. (B) Kaplan–Meier survival curves for three groups of mice followed up for 120 days after whole-thorax irradiation (UHDR-RT,30 Gy,473 Gy/s, n = 10; CONV-RT,30 Gy,0.07 Gy/s, n = 10; Control, n = 10). (C) Representative images of histopathological changes in lung samples (H&E staining, 72 h post-irradiation, scale bars: 100×, 200 μm, 400×, 50 μm; Masson staining, 120 days post-irradiation, scale bars: 100×, 200 μm, 400×, 50 μm). (D) Quantitative analysis of the number and area of alveoli and lung injury scores in H&E staining sections from the three groups. (E) Schematic diagram of whole-abdomen irradiation and observation indicators in mice. (F) Kaplan–Meier survival curves for three groups of mice followed up for 90 days after whole-abdomen irradiation (UHDR-RT,13 Gy,300 Gy/s, n = 10; CONV-RT,13 Gy,0.07 Gy/s, n = 10; Control, n = 10). (G) Representative images of histopathological changes in intestinal samples (H&E staining, 72 h post-irradiation, scale bars: 12.5×, 250 μm, 200×, 100 μm; Masson staining, 90 days post-irradiation, scale bars: 12.5×, 250 μm, 200×, 100 μm). (H) Quantitative analysis of the number of crypts per millimeter, villus height, and intestinal injury scores in H&E staining sections from the three groups. *p < 0.05, ** p < 0.01, ***p < 0.001, ****p < 0.0001 UHDR-RT reduced radiation-induced skin injury Mice subjected to skin irradiation were exposed 24 Gy. Skin damage first appeared 14 days post-irradiation, characterized by small areas of moist desquamation that gradually increased in size. By 60 days post-irradiation, skin thickening was observed (Fig. 4A). These phenomena were observed in all mice in the CONV-RT group, whereas no skin damage was observed in the UHDR-RT group. Dynamic skin damage scoring was performed for all mice (Fig. 4D), with higher scores indicating more severe damage. Results showed that damage in the CONV-RT group was significantly more severe than that in the UHDR-RT and Control groups (P < 0.01). The incidence of moist desquamation was as follows: Control group, 0%; UHDR-RT, 0%; CONV-RT group, 100% (Fig. 4E), with statistically significant differences observed. Two months post-irradiation, H&E staining (Fig. 4B) and Masson staining (Fig. 4C) of mouse skin tissue sections revealed no significant damage in the UHDR-RT and Control groups, whereas the CONV-RT group exhibited significant thickening of the epidermis (Fig. 4F: Control,11.47 ± 2.32 μm; UHDR-RT,13.23 ± 0.35 μm; CONV-RT,173.67 ± 23.07 μm), dermis (Fig. 4G: Control,142.33 ± 11.68 μm; UHDR-RT,151.00 ± 23.64 μm; CONV-RT,529.00 ± 236.29 μm), and increased fibrosis (Fig. 4H: Control,34% ± 14%; UHDR-RT,42% ± 3%; CONV-RT,53% ± 6%). At the end of the observation period, whole blood was collected for routine blood tests. Results indicated that 24 Gy CONV-RT radiation caused an increase in the percentage of neutrophils (NE%). Fig. 4Skin injury post right leg irradiation. (A) Photographs of the irradiated skin areas on the right lower limb of mice at 14 days, 30 days, and 60 days post-irradiation (UHDR-RT,24 Gy,250 Gy/s, n = 8; CONV-RT,24 Gy,0.07 Gy/s, n = 8; Control, n = 8). (B) Representative images of histopathological changes in skin samples stained with H&E 2 months post-irradiation (scale bars: 200×, 100 μm). (C) Representative images of histopathological changes in skin samples stained with Masson 2 months post-irradiation (scale bars: 200×, 100 μm). (D) Quantitative comparison of skin damage scores. (E) Prevalence of moist desquamation among different groups. (F) Quantitative analysis of epidermal thickness. (G) Quantitative analysis of dermal thickness. (H) Quantitative analysis of relative collagen per area. (I) Comparison of complete blood count parameters. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 Skin injury post right leg irradiation. (A) Photographs of the irradiated skin areas on the right lower limb of mice at 14 days, 30 days, and 60 days post-irradiation (UHDR-RT,24 Gy,250 Gy/s, n = 8; CONV-RT,24 Gy,0.07 Gy/s, n = 8; Control, n = 8). (B) Representative images of histopathological changes in skin samples stained with H&E 2 months post-irradiation (scale bars: 200×, 100 μm). (C) Representative images of histopathological changes in skin samples stained with Masson 2 months post-irradiation (scale bars: 200×, 100 μm). (D) Quantitative comparison of skin damage scores. (E) Prevalence of moist desquamation among different groups. (F) Quantitative analysis of epidermal thickness. (G) Quantitative analysis of dermal thickness. (H) Quantitative analysis of relative collagen per area. (I) Comparison of complete blood count parameters. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 UHDR-RT demonstrated comparable tumor control efficiency to CONV-RT To evaluate the tumor control effect, we established mouse tumor models using three clinically relevant tumor types for intraoperative radiotherapy: breast cancer (4T1), glioma (U87-MG), and pancreatic cancer (PAN02). When the tumor volume reached about 100 mm³, radiotherapy was administered. The schematic diagram of the irradiation dose, irradiation time, and observation indicators for each tumor-bearing mouse is presented in Fig. 5A. The tumor control effects of the CONV-RT and UHDR-RT groups were similar, with no statistical significance (P > 0.05). In the 4T1 and U87-MG tumor models, approximately 30 days post-irradiation, the tumor volume in the Control group (4T1:1935.98 ± 898.02 mm³; U87-MG:1711.87 ± 797.71 mm³) was significantly larger than that in the UHDR-RT group (4T1:816.79 ± 744.84 mm³; U87-MG:463.13 ± 121.21 mm³) and the CONV-RT group (4T1: 816.79 ± 744.84 mm³; U87-MG:463.13 ± 121.21 mm³) (P < 0.0001), while the tumor volume between the UHDR-RT and CONV-RT groups was similar (P > 0.05) (Fig. 5B). There was no significant difference in body weight changes among the three groups during the observation period (Fig. 5C). After tumor tissue was harvested, it was photographed (Fig. 5D) and weighed (Fig. 5E), which further confirmed that tumors in the Control group were significantly heavier than those in the UHDR-RT and CONV-RT groups. In the PAN02 tumor model, 25 days post-irradiation, the tumor volume in the Control group (2325.64 ± 755.05 mm³) was significantly larger than that in the UHDR-RT group (1025 ± 996.54 mm³) and the CONV-RT group (1210.75 ± 733 mm³) (P < 0.0001), while the tumor volume between the UHDR-RT and CONV-RT groups was similar (P > 0.05) (Fig. 5B). There was no significant difference in body weight changes among the three groups during the observation period (Fig. 5C). After tumor tissue was harvested, it was photographed (Fig. 5D) and weighed (Fig. 5E), which further confirmed that tumors in the Control group were significantly heavier than those in the UHDR-RT and CONV-RT groups (P < 0.0001). Fig. 5Tumor control effects of UHDR-RT and CONV-RT. (A) Schematic diagram of irradiation and observation indicators in three tumor models (breast cancer [4T1], glioma [U87-MG], and pancreatic cancer [PAN02]). (B) Tumor volume growth curves for UHDR-RT and CONV-RT groups (UHDR-RT: 4T1, 10 Gy, 192 Gy/s, n = 7; U87-MG, 10 Gy, 192 Gy/s, n = 9; PAN02, 16 Gy, 192 Gy/s, n = 8; CONV-RT: 4T1, 10 Gy, 0.07 Gy/s, n = 7; U87-MG, 10 Gy, 0.07 Gy/s, n = 9; PAN02, 16 Gy, 0.07 Gy/s, n = 8; Control: sham radiotherapy, 4T1, n = 7; U87-MG, n = 9; PAN02, n = 8). (C) Body weight changes in mice over the observation period. (D) Representative images of tumor tissues collected at the observation endpoint. (E) Tumor weights at the observation endpoint. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 Tumor control effects of UHDR-RT and CONV-RT. (A) Schematic diagram of irradiation and observation indicators in three tumor models (breast cancer [4T1], glioma [U87-MG], and pancreatic cancer [PAN02]). (B) Tumor volume growth curves for UHDR-RT and CONV-RT groups (UHDR-RT: 4T1, 10 Gy, 192 Gy/s, n = 7; U87-MG, 10 Gy, 192 Gy/s, n = 9; PAN02, 16 Gy, 192 Gy/s, n = 8; CONV-RT: 4T1, 10 Gy, 0.07 Gy/s, n = 7; U87-MG, 10 Gy, 0.07 Gy/s, n = 9; PAN02, 16 Gy, 0.07 Gy/s, n = 8; Control: sham radiotherapy, 4T1, n = 7; U87-MG, n = 9; PAN02, n = 8). (C) Body weight changes in mice over the observation period. (D) Representative images of tumor tissues collected at the observation endpoint. (E) Tumor weights at the observation endpoint. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 At the endpoint, tumor tissues from all three tumor models were subjected to H&E staining, immunohistochemical (IHC) staining, and immunofluorescence (IF) staining. H&E staining (Fig. 6A) revealed normal cell morphology in the Control group, whereas the UHDR-RT and CONV-RT groups exhibited extensive tumor cell death aggregates. The TUNEL (Fig. 6B and E) and γH2Ax (Fig. 6C and F) immunostaining results showed that the number of positive cells and the percentage of stained area were significantly higher in the UHDR-RT and CONV-RT groups compared to the Control group. In contrast, Ki-67 expression (Fig. 6D and G) was highest in the Control group, with statistically significant differences observed (P < 0.05). Fig. 6Histological examination and comparative analysis of tumors: H&E staining, immunohistochemical (IHC) staining, and immunofluorescence (IF) staining of tumor tissues, with quantitative analysis. (A) H&E staining of tumor tissues (scale bars: 400×,50 μm). (B) Tunel IF staining of tumor tissues (scale bars: 400×,20 μm). (C) γH2Ax IHC staining of tumor tissues (scale bars: 200×,100 μm). (D) Ki-67 immunofluorescence staining of tumor tissues(scale bars: 400×,20 μm). (E) Quantitative analysis of Tunel staining. (F) Quantitative analysis of γH2Ax staining. (G) Quantitative analysis of Ki-67 staining. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 Histological examination and comparative analysis of tumors: H&E staining, immunohistochemical (IHC) staining, and immunofluorescence (IF) staining of tumor tissues, with quantitative analysis. (A) H&E staining of tumor tissues (scale bars: 400×,50 μm). (B) Tunel IF staining of tumor tissues (scale bars: 400×,20 μm). (C) γH2Ax IHC staining of tumor tissues (scale bars: 200×,100 μm). (D) Ki-67 immunofluorescence staining of tumor tissues(scale bars: 400×,20 μm). (E) Quantitative analysis of Tunel staining. (F) Quantitative analysis of γH2Ax staining. (G) Quantitative analysis of Ki-67 staining. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 Discussion The CPIO-EFLASH device in this study was designed primarily to integrate electronic FLASH radiotherapy technology into the clinical application of IORT. The device has a minimum height of 3.2 m and occupies an area of at least 36 m². It features a superconducting linear accelerator with SSD of 50 cm and a dose rate of ≥ 80 Gy/s. For the first time, we validated the tumor control efficacy of this device in preclinical models, as well as its tissue-sparing capabilities in the brain, lungs, intestines, and skin. Specifically, the whole-brain irradiation was aimed at evaluating the brain-sparing efficacy in glioma radiotherapy; the whole-thoracic and skin irradiation were intended to assess the lung and skin-sparing efficacy in breast cancer radiotherapy; and the abdominal irradiation was designed to evaluate the intestinal tissue-sparing efficacy in pancreatic cancer radiotherapy. In recent years, the rapid advancement of IORT devices and intraoperative anesthesia management techniques has transformed IORT into an effective treatment modality for various types of tumors. Breast cancer, glioblastoma, and pancreatic cancer are common malignancies that can be treated with IORT in clinical practice. Therefore, we developed mouse models for these three tumor types and demonstrated that the CPIO-EFLASH device can effectively control tumor growth and achieve tumor control comparable to that of conventional dose-rate electron beams. Our findings are consistent with previous reports, showing that UHDR-RT and CONV-RT exhibit similar cytotoxic efficacy against tumor cells at the same dose. Favaudon et al. [11] compared the responses of lung cancer, breast cancer, glioma in mouse models to CONV-RT and UHDR-RT. They found no statistically significant difference in tumor control rates at the same dose. In an ovarian cancer mouse model, analysis of the number of tumor nodules and total tumor weight revealed that UHDR-RT exhibited similar tumor control efficacy to CONV-RT (P > 0.05) [20]. In the pancreatic cancer mouse model, proton radiotherapy reduced tumor recurrence, but no significant differences were observed between CONV-RT and UHDR-RT [27]. Additionally, studies in higher mammals have been conducted. In a study on feline nasopharyngeal carcinoma, normal tissues demonstrated good tolerance to high-dose single-fraction UHDR-RT while maintaining a high tumor control rate [13]. Some researchers have also suggested that UHDR-RT may have synergistic potential when combined with immunotherapy. UHDR-RT enhances T-cell infiltration within tumors at early time points and augments the efficacy of anti-programmed death-1 antibody (anti-PD-1 antibody, αPD-1) treatment [28]. These findings provide a strong theoretical basis for the future application of UHDR-RT in human cancer patients. In our study, two months after whole-brain irradiation (10 Gy,250 Gy/s), mice in the UHDR-RT group exhibited better spatial learning and memory abilities compared to those in the CONV-RT group. Several recent studies have reported the neuroprotective effects of UHDR-RT. Whole-brain irradiation at 10 Gy with an X-ray dose rate of > 100 Gy/s has been shown to reduce hippocampal cell division impairment and gliosis without causing significant anxiety, depression, or deficits in memory and learning functions [29]. Montay-Gruel et al. [30] demonstrated that UHDR-RT with large fractionation (3 × 10 Gy) effectively avoided radiation-induced cognitive deficits in learning and memory while maintaining tumor therapeutic efficacy. UHDR-RT outperformed CONV-RT in both short-term and long-term adverse central nervous system reactions, with significant benefits ranging from synaptic integrity to cognition and microvasculature preservation. These effects may be related to reduced inflammatory cytokine production and preservation of cerebral microvascular integrity [12, 30, 31]. For whole-thorax irradiation (30 Gy, 473 Gy/s), the CPIO-EFLASH device reduced acute lung injury and improved survival rates (P < 0.05), consistent with the previously reported lung-protective efficacy of high-energy X-ray UHDR-RT on our PARTER platform [16]. Chronic lung injury was observed four months post-irradiation, with the UHDR-RT group exhibiting less pulmonary fibrosis compared to the CONV-RT group. However, this observation did not reach statistical significance due to the low survival rate in the CONV-RT group (only one mouse survived at the endpoint of observation). Other study has shown that CONV-RT at a single dose of 15 Gy could induce pulmonary fibrosis in mice eight weeks post-treatment [11]. In contrast, UHDR-RT (≥ 500 Gy/s, ≤ 40 ms) did not induce pulmonary fibrosis associated with TGF-β cascade activation at the same dose. Pulmonary fibrosis was only observed at UHDR-RT doses above 30 Gy, with lower levels of pulmonary microvascular injury and bronchiolar epithelial cell apoptosis. The underlying mechanisms may involve reduced DNA damage and telomerase injury in normal lung tissue, thereby inhibiting cellular senescence [32, 33]. The gastrointestinal tract, characterized by rapidly proliferating and differentiating cells, is one of the most radiosensitive organs in the body. High-dose total body irradiation or clinical radiotherapy for abdominal and pelvic tumors can induce varying degrees of radiation enteropathy. In our study,90 days after whole-abdomen irradiation (13 Gy,300 Gy/s), the survival rate in the UHDR-RT group was significantly higher than that in the CONV-RT group. UHDR-RT also reduced acute intestinal injury, as evidenced by a higher number of intestinal crypts and greater villus height. Chronic intestinal injury was assessed three months post-irradiation, with neither group exhibiting intestinal fibrosis. However, this finding also lacked statistical significance due to the low survival rate in the CONV-RT group (only one mouse survived at the endpoint of observation). Diffenderfer et al. [27] reported that abdominal UHDR-RT reduced the loss of proliferating cells in the crypts during the acute phase and preserved regenerative crypts. Furthermore, UHDR-RT significantly decreased complete intestinal fibrosis. Ruan et al. [34] evaluated changes in gut microbiota composition following whole-abdomen UHDR-RT or CONV-RT in mice, finding that UHDR-RT induced fewer alterations in the gut microbiota, thereby maintaining intestinal homeostasis and attenuating intestinal inflammation. Additionally, our previous studies demonstrated that FLASH irradiation is less efficient in stimulating cytoplasmic double-stranded DNA (dsDNA) and activating the cGAS (cyclic GMP-AMP synthase) pathway in intestinal crypts compared to conventional X-rays. Under PD-L1 blockade, FLASH X-rays mitigate murine enteritis by alleviating harmful CD8 + T cell-mediated immune responses [35]. Regarding skin injury, we found that UHDR-RT not only reduced the incidence of radiation dermatitis in mice compared but also alleviated the severity of skin fibrosis eight weeks post-irradiation (24 Gy,250 Gy/s). A clinical trial in canine patients demonstrated that normal skin tissue exhibited good tolerance to single high-dose UHDR-RT (30–35 Gy) [36]. Similarly, in higher mammals (cats and pigs), UHDR-RT mitigated both acute and long-term adverse skin reactions. Conventional dose-rate irradiation at 25 Gy led to long-term (> 6 months) loss of hair follicles in pig skin, whereas UHDR-RT achieved this effect only at a total dose of 34 Gy [13]. In patients with cutaneous T-cell lymphoma, no acute or severe late skin toxicities were observed following UHDR-RT, indicating that UHDR-RT is safe and effective for normal skin [22, 37]. The skin-sparing mechanism of UHDR-RT may be related to the reduction of TGF-β1 and CXCL1 production, thereby minimizing radiation-induced skin toxicity [38]. Additionally, UHDR-RT preserves the level of epidermal stem cells and reduces hair follicle damage [13, 14]. The actual dose rate range adopted in our irradiation studies on lung, intestine, skin, and tumors was not constant, but all were significantly higher than 40 Gy/s, meeting the criteria for ultra-high dose rate irradiation. This study mainly focused on the dose factor, and the dose rate did not affect the reliability of the experimental results. In previous UHDR-RT studies using high-energy X-rays conducted by our research team, different dose rates (244–340 Gy/s) were also used for irradiating different tissues (throax, abdomen, tumor, et al.), which may similarly induce the FLASH effect (Lin B, et al). In recent years, several studies have reported negative results of ultra-high dose rate irradiation [39–41]. To further evaluate the impact of the dose rate parameter on the FLASH effect, we previously conducted a literature review and found that an average dose rate of 60 Gy/s or higher is required to induce the FLASH effect in brain tissue, an average dose rate of ≥40 Gy/s seems to be the threshold for inducing the FLASH effect in mouse lung tissue, and the minimum average dose rate for inducing the FLASH effect in skin is 65 Gy/s [42]. However, the average dose rate is not the only factor affecting the FLASH effect; overall, an average dose rate of 40 Gy/s seems to be the minimum dose threshold for triggering this effect. Single or multiple pulses, as well as the characteristic parameters of pulse structure (including single pulse dose, pulse width, pulse interval, pulse frequency, and total irradiation time), have also been found to influence the FLASH effect. Additionally, to clarify whether CPIO-EFLASH can trigger a protective effect on abdomen, we intentionally set the irradiation dose of UHDR-RT to be slightly higher than that of CONV-RT, yet a significant FLASH effect was still induced under this condition. Furthermore, we reviewed classic literature on FLASH radiotherapy, where the researchers also observed the protective effect of FLASH on normal tissues by setting the UHDR-RT irradiation dose higher than that of CONV-RT [11, 16]. To our knowledge, a LIAC FLASH system (S.I.T. Sordina IORT Technologies) is currently under development [43]. Existing studies have primarily utilized the FLASH Mobetron LINAC (IntraOp, Sunnyvale, CA, USA), which has demonstrated abdominal sparing effects of FLASH-RT in murine models [44]. Sesink A. et al. [45] further reported that reducing the dose per fraction attenuates the normal-tissue sparing effect of UHDR-RT, providing key biological evidence to guide optimization of fractionation strategies. In addition, UHDR-RT has been explored clinically in patients with progressive cutaneous metastatic melanoma [46], in whom escalating radiation dose did not lead to increased acute skin reactions. Collectively, electron FLASH irradiation can markedly reduce normal tissue toxicity while preserving antitumor efficacy, with good reproducibility across different centers [44]. Our work further expands the preclinical application of electron FLASH in IORT settings: using the CPIO-EFLASH platform, we confirmed robust antitumor activity in glioblastoma, pancreatic cancer, and breast cancer models, and systematically verified multi-organ protection (brain, lung, intestine, and skin). These promising results have encouraged us to advance toward clinical translation, and a Phase I clinical trial is planned for initiation in 2025 (ChiCTR2500106390). We anticipate reporting the outcomes of its clinical application in humans in the future. This study also has several limitations. First, we used mouse subcutaneous tumor models rather than orthotopic tumor models or models simulating the IORT scenario to validate the tumor control efficacy of the CPIO-EFLASH device. Second, the insufficient number of surviving mice in the CONV-RT group compromised the statistical significance of the chronic lung and intestinal injury assessment. Despite these limitations, this study sufficiently demonstrated that UHDR-RT can reduce radiation-induced damage to brain, lung, intestinal, and skin tissues, highlighting its potential to protect normal tissues. Conclusion The CPIO-EFLASH can achieve ultra-high dose rates (≥ 40 Gy/s at an SSD of 50 cm) and trigger significant normal tissue-sparing effects. Integrating electronic FLASH technology into intraoperative radiotherapy may bring potential clinical benefits to patients by effectively treating tumors, while minimizing radiation-induced injury to normal tissues. Our findings highlight the necessity for further clinical trials of CPIO-EFLASH in intraoperative radiotherapy. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1