Radiation therapy (RT) is employed in roughly one-half of all cancer cases and contributes to about 40% of curative treatments worldwide (1). The recent arrival of ultra-high-dose-rate (UHDR), RT has reinvigorated translational efforts to increase the therapeutic ratio by preferentially sparing normal tissue while maintaining tumor control (“FLASH” sparing) (2–4). Elegant proof-of-principle experiments in skin, lung and brain have demonstrated dramatic reductions in toxicity when doses are delivered in micro-second pulses at ≥ 30 Gy/s compared with conventional dose rates (CDR) (5–9). Yet extension of these findings to the abdomen—a site where acute gastrointestinal (GI) syndrome remains a foremost dose-limiting toxicity—has been inconsistent. Whole-abdomen electron FLASH protects crypt stem cells and improves survival in some reports, whereas pencil-beam-scanned proton FLASH and synchrotron proton beams have produced neutral or even worse outcomes relative to CDR irradiation (10, 11). A potential explanation is that the FLASH effect is particularly sensitive to biological context and experimental nuance that is not being controlled for across studies.

A unifying hypothesis for FLASH sparing invokes transient oxygen depletion: extremely intense pulses have been postulated to consume dissolved O₂ faster than it can be replenished, driving tissue into a radioprotective hypoxic state for the brief period in which radical-mediated damage is fixed (12–15). Bulk tissue measurements have refuted this idea through systematic study of skin, and tumor, however still debate exists on whether there could be regional or micro-localized hypoxia induced during the irradiation in FLASH. The data also seems to be dependent upon the total dose as well as the tissue oxygen, and so there seems to be both a minimum and a maximum dose threshold to see the FLASH effect (16, 17). It is therefore likely that oxygen may be a factor in this complex range where FLASH tissue sparing can be seen. If it is true that oxygenation level affects the observation, then any variable that alters baseline or dynamic tissue oxygenation could modulate—or even mask—the FLASH effect. Two variables stand out in small-animal work: the choice of anesthetic and the composition of the carrier or supplemental gas.

Intraperitoneal ketamine/xylazine (K/X) and inhaled isoflurane dominate murine RT studies because they are facile and inexpensive. Unfortunately, they create very different physiological milieus. K/X produces profound bradycardia, respiratory depression, and peripheral vasoconstriction, leading to arterial hypoxemia and tissue pO₂ values as low as 15 mmHg in skin and brain. Isoflurane, in contrast, raises heart rate, increases tidal volume, and vasodilates vessels; when delivered in room air it maintains skin pO₂ near 25–30 mmHg, and when delivered in 100% oxygen it can drive tissue pO₂ above 50 mmHg. Carrier-gas oxygenation matters independently of the volatile agent: supplemental oxygenation during irradiation sharply increases murine skin P_O2 values and negates the FLASH effect (18–20).

Despite these well-documented effects, anesthetic protocols are rarely reported in pre-clinical FLASH papers, and when they are, oxygen supplementation is common. In a recent systematic review of electron FLASH in vivo studies, fewer than one-third specified the fraction of inspired oxygen, and none quantified tissue pO₂. Recent results showed that this omission is not benign: in murine hind-leg skin, breathing 100% oxygen during isoflurane anesthesia abolished FLASH sparing, whereas room-air anesthesia preserved it; moreover, female mice—whose dermis exhibited higher pO₂ than males—ulcerated sooner after UHDR irradiation (18, 21). These data strengthened the link between oxygen tension and FLASH, but they also raised a critical question for abdominal irradiation: does the anesthetic itself, independent of inspired O₂, dictate whether gut tissues experience protection or injury at UHDR?

The small intestine is among the most radiosensitive organs, and weight-loss kinetics as well as crypt-regeneration assays are gold-standard read-outs for acute GI toxicity. Several groups have attempted to exploit FLASH in this setting, yet outcomes have ranged from substantial sparing to overt harm. None of these studies compared different anesthetic regimens head-to-head or measured intraperitoneal oxygenation. Given that intestinal perfusion is highly responsive to vasoactive and respiratory changes, and that K/X and isoflurane exert opposite effects on both, anesthetic selection may be a hidden confounder driving inter-study variability.

Here we directly address this gap by evaluating FLASH sparing in the mouse total abdominal irradiation model, anesthetized with either KX or isoflurane in room air. To mechanistically link any observed differences to oxygen availability, a parallel cohort received intravenous PdG4 Oxyphor and real-time phosphorescence lifetime imaging (OxyLED) was used to record intra-abdominal pO₂ immediately after laparotomy (22).

Using a 9 MeV electron beam, two previously established doses were delivered, at 14 or 16 Gy, with broad field irradiation to the abdomens of C57BL/6 mice under two commonly interchangeable anesthetics (10): (i) intraperitoneal ketamine/xylazine (K/X, 100/10 mg/kg), and (ii) inhaled isoflurane (ISO, 3% induction, 1.5% maintenance) in room air. Supplemental oxygen was deliberately avoided to isolate anesthetic-specific physiology. Employing time-to-25% weight-loss as an objective endpoint, we asked whether the magnitude of FLASH depended on the anesthetic type.

At the 14 Gy dose, a robust radioprotective effect was seen in UHDR mice anesthetized with K/X compared to ISO. Though a statistically significant sparing effect was not seen between K/X UHDR and CDR mice, it is apparent that this may have been caused by limited statistical power from the few events in the UHDR K/X arm. This is supported by a power analysis showing that a group size (n) of nearly 400 mice is needed to achieve 80% detection power (alpha = 0.05) using the baseline event rate (~0.12) and censoring rate (0.3) observed in the 14 Gy K/X UHDR arm. Regardless, at termination of study (14 days), 80% of UHDR K/X arm had not reached 25% weight loss, compared to 40% in the CDR K/X arm. Importantly though, there is no apparent FLASH effect abdominal sparing in mice anesthetized with ISO, indicating that with ISO anesthesia this would not be seen at this dose level.

At the increased 16 Gy dose, K/X administration showed notable radioprotection to the UHDR irradiated mice, as compared to the CDR mice. A significant sparing effect was seen when comparing UHDR to CDR irradiated mice under the K/X condition as well. At the termination of study, 40% of the UHDR K/X mice had not reached the 25% weight loss endpoint, in comparison to CDR mice who all reached the endpoint by 7 days. These results support the conclusion that ISO anesthesia negates the FLASH sparing effect at this dose level.

To investigate a potential mechanism for loss of sparing under ISO anesthesia, intraperitoneal oxygen measurements were performed using an injected optical oxygen reporter and fiber measurement system in six mice. A significant, nearly 2.5-fold difference in oxygen tension was seen between ISO (average ~62 ± 4 mmHg) and K/X (~26 ± 10 mmHg) conditions. Anesthesia affects tissue oxygenation in a variety of organ systems through multi-modal response including respiratory, cardiac, and vessel tone (perfusion) alteration. Still, the GI system, perfused by mesenteric and splanchnic vasculature, are specially responsive anesthetic effects (23). Considering this in the context of recently published work investigating FLASH sparing under varying oxygen conditions in mouse skin (13, 16, 18, 21), this data supports the interpretation that an oxygen-based mechanism affects the observed differences in radiosensitivity between ISO and K/X anesthesia. We see, again, that FLASH sparing is particularly sensitive to tissue oxygenation, beyond what would be expected from our conventional understanding of oxygen radiosensitization and the oxygen enhancement ratio.

There is now a wide body of literature demonstrating that tissue oxygenation is a key determinant of the magnitude of FLAH sparing across organs (5, 17, 20, 24). This data suggests that an oxygenation threshold or spectrum must exist, above which FLASH sparing is not significant. It is widely known that oxygen enhanced damage through peroxyl formation and DNA damage fixation, and so higher oxygen contributes to higher damage, but this is only seen in the range of local pO₂ <10 mmHg. Thus, it seems plausible that with UHDR irradiation that some local depletion occurs or local change in reactive oxygen species occur, which only significantly affects the normal tissues with lower initial pO₂. Questions remain about what this threshold is, whether it is organ-specific, and if it is modulated by total dose, dose-rate, or fractionation. Once these questions are answered, we can then tackle what we consider to be the key oxygenation question in translational UHDR: Are physiological oxygen values in awake, normally ventilating humans low enough to see a clinically meaningful FLASH sparing effect?

Our study is not without limitations. First, although we see a potent sparing in in mice who were anesthetized with K/X and received 14 Gy of UHDR radiation compared to CDR, we fail to detect a statistically significant difference using Kaplan-Meier (KM) analysis. This is attributed to the limited number of mice meeting the endpoint (2) in the UHDR arm, which affords very little statistical power to the log-rank test. Still, considering the two dose levels 14 Gy and 16 Gy together, FLASH sparing is most apparent under the K/X condition across doses. The second limitation is the inherent uncertainty in the abdominal compartment that is the primary driver of radiation toxicity/weight loss. In Total Abdominal Irradiation (TAI), numerous organs are irradiated with vastly different radiosensitivities. Though weight loss following TAI is typically attributed to small bowel damage, the more nuanced reality is that damage to many of the intrabdominal structures, aside from small bowel, can lead to weight loss and mortality. Although not quantified in this manuscript, significant hydronephrosis, liver damage, gastric enlargement, and enteritis were noted in post-mortem necropsy of mice that very likely contributed to weight loss. This leads to the third limitation, which is the lack of specifity in the abdominal compartment probed in our oxygen measurements. The wide-field oxygen measurements performed centered on the bowel to maximize signal from this organ of interest. This means that these oxygen values are an average of numerous abdominal structures within the irradiation field, rather than any one organ, but were weighted toward the bowel. Still, the results are clear: intrabdominal oxygen levels under ISO anesthesia are significantly higher than under K/X anesthesia. As oxygen measurement technologies advance, more organ-specific measurements would add value to TAI studies.

In summary, by explicitly integrating anesthetic physiology, dosimetry verified to within 3%, and direct oxygen measurements, the present work identified one of the persistent sources of variability relevant to pre-clinical FLASH research, which is anesthesia method. This is particularly important in the light of pre-clinical animal use refinement techniques moving away from injectables toward gaseous anesthesia. There needs to be practical guidance for experimental design as the field advances toward FLASH clinical trials. While direct measurement of tissue oxygen is challenging, future investigations should consider tighter control and reporting of experimental factors that can alter tissue oxygenation.