Oxygen Consumption In Vivo by Ultra-High Dose Rate Electron Irradiation Depends Upon Baseline Tissue Oxygenation

Abstract

Purpose:

This study aimed to assess the impact of tissue oxygen levels on transient oxygen consumption induced by ultra-high dose rate (UHDR) electron radiation in murine flank and to examine the effect of dose rate variations on this relationship.

Methods and Materials:

Real-time oximetry using the phosphorescence quenching method and Oxyphor PdG4 molecular probe was employed. Continuous measurements were taken during radiation delivery on a UHDR-capable Mobetron linear accelerator. Oxyphor PdG4 was administered into the subcutaneous tissue of the flank skin 1 hour before irradiation. Skin oxygen tension (pO₂) was manipulated by adjusting oxygen content in the inhaled gas mixture and/or by vasculature compression. A skin surface radiation dose of 19.8 ± 0.3 Gy was verified using a calibrated semiconductor diode dosimeter. Dose rate was varied across the UHDR range by changing linear accelerator cone length and pulse repetition frequency.

Results:

The decrease in pO₂ per unit dose during radiation delivery, termed oxygen consumption g-value (g_O2, mmHg/Gy), was significantly influenced by tissue oxygen levels in the range 0 to 65 mmHg under UHDR conditions. Within the 0 to 20 mmHg range, g_O2 exhibited a sharp increase with rising baseline pO₂, plateauing at 0.26 mmHg/Gy. Dose rate variations (mean values, 25–1170 Gy/s; per pulse doses of 2.5–9.8 Gy) were explored by varying both cone length and pulse repetition frequency (10–120 Hz) with no significant changes in g_O2. Conventional dose rate irradiation resulted in no discernible changes in pO₂.

Conclusions:

The results show significant differences in the radiation-chemical effects of UHDR radiation between hypoxic and well-oxygenated tissues. Similar trends between earlier published in vitro and in vivo experiments presented herein suggest the chemical mechanisms driving the dependencies of g_O2 on pO₂ are similar, potentially underpinning the FLASH effect. Importantly, significant variations in baseline pO₂ were observed in animals kept under identical conditions, underscoring the necessity to control and monitor tissue oxygen levels for preclinical investigations and future clinical applications of FLASH radiation therapy.

Introduction

Ultra-high dose rate (UHDR) radiation therapy (RT), termed FLASH-RT, has garnered considerable attention. Characterized by high delivery dose rates (usually >40 Gy/s), FLASH-RT has been shown to reduce normal tissue toxicity.¹ The mechanisms underlying this effect remain undetermined, but the existing experimental data suggest that oxygen may play a role. A recent study has shown that oxygen content in inspiration gas mixtures, and the resulting tissue oxygenation, may be predictive of expected outcomes in skin response studies with electron FLASH.² In the mid-1900s, multiple works were published on the reduction or elimination of high dose rate sparing effects using nitrogen gas to induce hypoxic conditions in vitro^3,4 and in vivo.⁵

The potential relationship between UHDR/FLASH and oxygen is related to the fact that tissue oxygenation is known to be one of the most significant modifiers of tissue damage by radiation. Oxygen-involving reactions occur downstream from the initial direct ionization of biomolecules and/or water radiolysis. In irradiated cells, oxygen covalently attaches to DNA-, protein-, and lipid-derived radicals, resulting in the oxygen fixation effect, which is quantitatively reflected in the oxygen enhancement ratio.⁶ The resulting peroxyl radicals cannot be converted back to the original biomolecules, but could, especially in the case of lipids, initiate chain reactions, leading to fixation of more oxygen and more damage. On the other hand, tissue oxygenation is one of the key environmental variables that can be different between healthy and tumor tissues. Solid tumors are frequently hypoxic (average oxygen partial pressure, pO₂ ≤ 10 mmHg) as compared with most healthy tissue (normal range pO₂ ≈ 20–40 mmHg).⁷ The timescale of FLASH delivery (10⁻⁶–10⁻² s) is similar to the timescales of a number of reactions involving oxygen and oxygen-derived species (Fig. 1).^6,8–11 At the same time, oxygen’s involvement and the radiation-chemical events are clearly upstream of biological responses, which require much longer times to develop. The nature of alteration in pO₂ upon UHDR delivery could be of high significance because of oxygen’s importance as a radiosensitizer and involvement in vital cellular processes and reactions.

Fig. 1.

Radiation delivery and chemical pathway timescales. (A) Timescales of UHDR radiation delivery and beam structure relative to observed rapid oxygen consumption and in vivo resupply or recovery of oxygen to irradiated tissue. (B) Timescales of water radiolysis cascade and secondary radical production.^8–11 Upside down triangles note lifetimes of hydrogen peroxide and superoxide.⁶ Reactions that consume oxygen are indicated with red dashed arrows. Abbreviations: PRF = pulse repetition frequency; UHDR = ultra-high dose rate.

Oxygen is a unique molecule that plays an important role in RT and can be measured in tissue in vivo from before to after the radiation pulse.^12–14 The amount of consumed oxygen (ΔpO₂) appears to increase with decreasing dose rate,^12,14 which could be related to a variation in the degree of oxygen fixation as the dose rate increases to FLASH-inducing levels. The oxygen consumption g-value (g_O2) is insufficient to induce complete depletion of oxygen to the levels of radiobiologically relevant hypoxia from a single clinical radiation dose.^12–16 It is also clear that the level of depletion of oxygen is affected by the composition of the medium,¹⁷ and it has been shown that the g_O2 values characteristic of the intracellular milieu are similar to those measured in model buffers and/or in vivo.¹⁵ The observable changes in g_O2-values have also been determined to be dependent on initial oxygen tension (pO₂) in model solutions,^{13,14,16,18,19} inside cells in vitro,¹⁵ and in vivo in the vascular and extracellular space,^12–14 implying that oxygen consumption and the FLASH effect may be linked, but the functional form of this dependence has not been fully elucidated in vivo. This work builds on recent measurements of radiation-induced oxygen consumption.^2,17,20–22 This study examined the dependence of oxygen consumption as a function of baseline oxygen level as well as on the dose rate and pulse repetition frequency (PRF) in a detailed manner in vivo.

Methods and Materials

Animals

All animal studies were approved by the local Insitutional Animal Care and Use Committee (IACUC), and all procedures followed the approved protocol. C57Bl/6J mice were used for the study with equal gender use (38 males and 34 females). These were supplied at 8 to 10 weeks of age from Jackson Laboratories, and they were allowed to acclimate for at least 1 week before beginning the experiment.

One hour before the planned irradiation, mice were given a 3-minute anesthesia induction with 3% isoflurane, and when fully unconscious the left leg and leg area were shaved. Oxyphor PdG4 (described in section “in vivo oximetry”) was then locally injected subcutaneously into the left flank. Mice were allowed to wake up for up to 1 hour after this and then in preparation for radiation delivery, the mice were anesthetized again using isoflurane (induction: 3% isoflurane delivered at 500 mL/min for 3 minutes, maintenance: 1.5% isoflurane delivered at 100 mL/min) in either room air, 100% oxygen, or carbogen. Anesthesia timing was previously analyzed by Tavakkoli et al,² and systematic variations in anesthesia and vasculature compression were specifically used to vary tissue pO₂ values for this study. Mouse body temperature was maintained using a heating pad below the mouse. Measurement of pO₂ was completed during irradiations, and mice were killed while unconscious after completion of each study.

Modulation of initial tissue pO₂

For the modulation of initial tissue oxygen tension, 4 cohorts of 8 mice each (4 male, 4 female) were established for each of the 2 UHDR irradiation geometries employed. Breathing conditions consisted of isoflurane mixed with either room air, 100% oxygen, or carbogen (95% O₂, 5% CO₂). A fourth cohort using isoflurane and room air with added vascular compression of the leg above the site of irradiation was included to attain local hypoxia in the region of irradiation. Compression was achieved through tying a piece of thread around the base of the leg, thereby restricting blood flow to the irradiation site without obscuring the delivery field. These conditions allowed for a wide range of initial pO₂ values of 0 to 65 mmHg, accounting for the full range of physiological oxygen tensions in tissue.⁷

Three mice (1 male, 2 female) were kept in a separate cohort for conventional dose rate delivery with room air breathing as a negative control group as no observable change in pO₂ was expected in accordance with previous studies.^12,14

Radiation delivery and dosimetry

A UHDR-capable Mobetron linear accelerator (IntraOp Inc) was used for all irradiation studies in mice, using a custom collimator to shape the beam to a 1.6-cm diameter circular field, which was centered on the left flank of each mouse. The collimator featured a 3D-printed polyethylene terephthalate glycol cone to allow for reproducible placement of a calibrated diode EDGE Detector (Sun Nuclear Corp) and the phosphorimeter fiber optic cable upstream of the positioned mouse (Fig. 2A). The fiber optic placement was designed such that the field of view coincided with the radiation delivery field to allow for simultaneous delivery and measurement without perturbing the delivered beam. Likewise, the diode dosimeter was placed in a lateral position with respect to the field.

Fig. 2.

Set-up, methodology, and beam structure. (A) Photo of mouse positioned for radiation delivery and measurements, with red light from phosphorimeter showing on mouse leg at site of subcutaneous (SQ) injection. (B) The table with all beam delivery parameters for the 2 UHDR radiation delivery geometries used, short cone and face plate. SSD, D_P, D_M, τ_IR, and Δt correspond to source-to-surface distance, peak dose rate, mean dose rate, total irradiation time, and interpulse time. (C) Example of beam control toroid trace showing UHDR pulse temporal structure, with full-width half-maximum of 3.07 μs. (D) Oxygen data over time, showing pO₂ values for 5 different radiation deliveries at varying initial tissue oxygen levels. Black points represent acquired oxygen values, and red lines represent the observed change in pO₂ (ie, ΔpO₂) due to and during the UHDR radiation delivery. Abbreviation: UHDR = ultra-high dose rate.

For dosimetry, the EDGE Detector was calibrated in a fixed lateral position under experimental geometry with a diamond detector prototype specifically designed for use in UHDR beams (flashDiamond) (PTW) at the center of the field to calibrate diode measurement to central axis dose. Diode readings were taken for each mouse delivery, then converted to skin surface dose using a calculated conversion factor. The EDGE Detector and flashDiamond were previously characterized by Rahman et al²³ and Tessonnier et al²⁴ for UHDR dosimetry use.

The Mobetron linear accelerator features interchangeable cones allowing for 3 cone lengths for collimator placement, termed the face plate, short cone, and long cone. Using the custom collimator, these 3 positions correspond to source to surface distances of 20.3, 37, and 46 cm. To evaluate dose rate effects on the impact of baseline tissue oxygenation on g_O2, 2 delivery geometries in UHDR mode were employed each with a unique source to surface distance and dose per pulse, whereas field size and pulse width were kept constant. The geometries utilized in this work, referred to in this work as the high dose per pulse (or face plate) condition and the low dose per pulse (or short cone) condition, combined with variation in PRF allowed for direct comparison of several beam structures that fall within the UHDR regime. See Figure 2B for a more detailed parametric comparison of the 2 conditions.

Radiation deliveries consisted of 3 μs pulses (9.79 ± 0.08 Gy/pulse at the face plate and 2.50 ± 0.02 Gy/pulse at the short cone). For the high dose per pulse condition, instantaneous, or intrapulse, dose rate was 3.26 × 10⁶ Gy/s. For the low dose per pulse condition, instantaneous dose rate was 8.34 × 10⁵ Gy/s. Temporal pulse structure was verified and monitored using the beam control toroids in the Mobetron head (Fig. 2C).

In vivo delivery of a target dose of 19.8 ± 0.3 Gy was achieved to the skin surface with the 9 MeV UHDR electron beam using 2 or 8 pulses at the high and low dose per pulse conditions, respectively. Mice were irradiated twice while under anesthesia, PRFs of 120 and 10 Hz. The irradiation order was counter-balanced within the cohort and found to have no effect on g_O2 versus initial pO₂. Moreover, oxygen levels were monitored for several minutes between irradiations to ensure adequate time for recovery and stabilization before the second exposure. For conventionally irradiated mice, a target dose of 19.6 ± 0.3 Gy to the skin surface was delivered with a 9 MeV electron beam at a total delivery time of 123 seconds, corresponding to a time-averaged mean dose rate of 0.16 Gy/s. Conventional deliveries on the Mobetron have a fixed pulse width of 1.2 μs and a fixed PRF of 30 Hz.

In vivo oximetry

Oxyphor PdG4 (or Oxyphor G4) was used as an oxygen probe for real-time oximetry (Oxygen Enterprises) by the phosphorescence quenching method. PdG4 has been well characterized for in vitro and in vivo use, making it a robust and translatable probe for oxygen sensing.²⁵

In this work, 80 μL of 20 μM PdG4 in phosphate buffer solution was injected subcutaneously in the mouse flank one hour before radiation delivery, allowing for time to equilibrate and diffuse within the tissue, as previously established.^2,26

An Oxyled phosphorimeter (Oxygen Enterprises) was used to measure the phosphorescence lifetimes that allow direct estimation of in vivo oxygen tension throughout radiation delivery. The measurements were performed at a sampling rate of 5 Hz. The phosphorimeter system is based on an avalanche photodiode detector that recorded the phosphorescent emission, with excitation via a fiber-coupled pulsed LEDs (635 nm). The fibers were kept in a fixed position ~15 mm from the center of the radiation field on the mouse to avoid perturbation of the radiation field. The Oxyled device and logging computer were kept approximately 1 m from the radiation field. Local pO₂ was calculated from the measured phosphorescent lifetime using previously determined calibration parameters.²⁵

Real-time, continuous acquisition of oxygen data allowed for determination of change in oxygen directly due to delivery of radiation (Fig. 2D). Initial or baseline pO₂ was determined by averaging several predelivery data points. This value was then subtracted from postdelivery measurements to calculate the change in oxygen partial pressure. All ΔpO₂ values were normalized to delivered dose and are reported as oxygen consumption g-values, g_O2, in mmHg per unit dose (mmHg/Gy).

Statistical analysis and functional fit

Each measurement was completed on one animal in one oxygenation condition, and repeated measures of a single initial pO₂ value was not possible given the high variability in baseline values. Error bars for each value of ΔpO₂ were estimated by the error in fitting of the data types as shown in Figure 2D. Upon dose-normalization for the calculation of g_O2, the error of each fit and the uncertainty of the semiconductor diode dosimeter²³ were added in quadrature to produce the final error bar for each measurement. The final g_O2-values versus initial pO₂ were fitted to an arbitrary equation y = c − a*exp(−b*x), that appeared to visually follow the measured trends, where y was g_O2 and x was initial pO₂. Notably, the chosen model accommodates 2 critical trends observed in the experimental results: the presence of a saturation point, indicating that g_O2 reaches a maximum value beyond which it does not increase despite further changes in initial pO₂, and intersection with the origin, reflecting that there is no oxygen consumption when tissue pO₂ is zero. The baseline term, c, corresponded to the g_O2-value of normoxic tissue. These empirical fits were added in Figures 3 to 5 as visual aids to represent and compare observed data trends under varied beam structures and geometric conditions.

Fig. 3.

Dose linearity. Dose response for in vivo ΔpO₂ relative to 25–35 mmHg baseline oxygen at 10 dose levels, ranging from 2.4 to 24 Gy in increments of 2.4 Gy. Empirical fit of functional form ΔpO₂ = m*Dose, where ΔpO₂ is in mmHg and dose is in Gy. The slope was found to be 0.24 ± 0.01 mmHg/Gy.

Fig. 5.

Pulse repetition frequency comparison. In vivo ΔpO₂ measurements shown by pulse repetition frequency (PRF) at 10 and 120 Hz. The inlaid table shows fit parameters with 95% confidence intervals using the functional form: y = c − a*exp (−b*x). (A) Data set of mice irradiated at higher UHDR, having mean dose rates 1170 and 98 Gy/s, respectively, at 120 and 10 Hz PRF. (B) Data set of mice irradiated at lower UHDR, having mean dose rates of 300 and 25 Gy/s, respectively, at 120 and 10 Hz PRF. Abbreviation: UHDR = ultra-high dose rate.

Results

Dependence of the oxygen consumption g-value (g_O2) on baseline tissue oxygenation

For proper analysis and comparison of the g_O2 metric across variations in delivered dose, a dose linearity investigation was performed (Fig. 3). Five male mice breathing room air were irradiated at multiple dose levels with the low dose per pulse geometry. Mice were allowed to recover to baseline pO₂ before sequential irradiations and killed immediately after radiation delivery. Only male mice were used because baseline oxygenation in female animals was below 20 mmHg, deemed as the approximate cutoff for baseline pO₂ dependence of g_O2 (Fig. 4), whereas the male mice had baseline oxygenation between 25 and 35 mmHg. It was found that ΔpO₂ changes linearly with dose for the investigated range (1–10 pulses, corresponding to 2.4–24 Gy) with an R² value of 0.99 (see Fig. E1 for dose linearity residuals plot), and hence the g_O2 value was nearly constant.

Fig. 4.

Oxygen consumption by oxygen condition. In vivo ΔpO₂ measurements of mice irradiated at high dose per pulse (A) and at low dose per pulse (B) plotted versus initial tissue pO₂. Data points are labeled by mouse condition: breathing carbogen gas, 100% O₂, room air, and room air with vascular compression, labeled by color with magenta, black, blue, and red, respectively. The empirical fit line had functional form y = c − a*exp(—b*x). (A) For the 9.8 Gy/pulse cohort, a = 0.29 mmHg/Gy, b = 0.12 mmHg⁻¹, and c = 0.27 mmHg/Gy. (B) For the 2.5 Gy/pulse cohort, a = 0.29 mmHg/Gy, b = 0.14 mmHg⁻¹, and c = 0.25 mmHg/Gy. The inlaid plot shows distribution of initial tissue oxygenation for each breathing condition.

The dependence of oxygen consumption in vivo on the tissue oxygenation was investigated across a range of initial pO₂ values, 0 to 65 mmHg. Radiation-induced changes in pO₂ were quantified, normalized to the delivered dose, and plotted against initial tissue pO₂ for the 2 delivery geometries (Fig. 4). In both geometries, the plots were similar, each featuring 2 distinct regions. At lower oxygen levels, from 0 to 20 mmHg, g_O2 was found to increase sharply with an increase in baseline pO₂. After this sharp increase, g_O2 values plateaued at 0.27 and 0.25 mmHg/Gy for mice irradiated at high and low dose per pulse conditions, respectively. This trend is consistent with previously published in vitro results evaluating oxygen dependence of oxygen consumption in model solutions for both electron and proton beamlines^{13,14,16,18,19} and inside cells.¹⁵

In previous studies, oxygen consumption above 20 mmHg was found to be higher for conventional than FLASH dose rates.^{13,14,16,18,19} Additionally, it was found in our previous work that for the same intrapulse dose rate, oxygen consumption in vitro in model solutions was higher for lower time-averaged dose rates (ie, lower PRFs).¹⁶ As stated in section “Radiation delivery and dosimetry,” 2 delivery geometries were employed, and each mouse was irradiated multiple times using PRFs of 10 and 120 Hz to assess whether modulating time-averaged dose rate would have an effect on the baseline pO₂ dependence in vivo. It was found that g_O2 values were the same for the PRFs examined in either geometry (Fig. 5).

Upon comparison of the trends from the 2 geometric conditions, no significant difference was observed between the 2 empirical fits, as the fit parameters for each data set fell within the 95% confidence intervals of the other (Fig. 6). Combined with the results shown in Figure 5, these findings reveal that unlike in model solutions, tissue in vivo g_O2 values do not depend on PRF in the UHDR regime for the frequencies tested (10 and 120 Hz).

Fig. 6.

Dose per pulse comparison. Comparison of g_O2-values for the high dose per pulse (green) and low dose per pulse (yellow) conditions. The inlaid table shows fit parameters with 95% confidence intervals using the functional form: y = c − a*exp(−b*x). Shaded regions represent the 95% confidence intervals for each of the 2 empirical fits.

Oxygen measurements were also taken for mice under conventional dose rate delivery conditions and room air breathing. As expected, based on previous reports,^12,14 no apparent changes in oxygenation were detected (Fig. 7) due to the fast resupply of oxygen to the irradiated tissue outpacing the radiation-induced consumption of oxygen for conventional dose rates (0.16 Gy/s).

Fig. 7.

Conventional dose rate comparison. Box plots showing in vivo ΔpO₂ measurements by cohort including conventional dose rate (CDR) data in orange. CDR data reflects no observed ΔpO₂. (A) Comparison of CDR data to the high dose per pulse data set. (B) Comparison of CDR data to the low dose per pulse data set.

Discussion

This work is the first comprehensive study investigating the dependence of oxygen consumption g-values (g_O2) on tissue pO₂ over a broad range of physiologically relevant oxygen concentrations in vivo, thus significantly expanding the data obtained previously.^12–14 It was found that g_O2 values plateau at 0.26 ± 0.01 mmHg/Gy for oxygen tensions above 20 mmHg for the combined data set of all 132 deliveries (Fig. E3). The consumption of oxygen decreases systematically below this level down to 0, with the 50% crossover point at approximately 8 mmHg. This trend resembles that seen in vitro when measurements were performed using protein solutions,^13,16 model buffers,^14,15 or in vitro inside cells.¹⁵

Importantly, we also found that changes in PRF did not affect g_O2-values across the full range of baseline pO₂ values, which contrasts our earlier results for model solutions,¹⁶ where a clear trend was observed showing that oxygen consumption decreases with an increase in PRF for the same intrapulse dose rate. The range of dose rates tested in this study was necessarily limited to UHDRs (25–1170 Gy/s), because at lower dose rates, g_O2 measurements were obstructed by the rapid resupply of oxygen by diffusion from capillaries in tissue.

A major microenvironmental distinction between solid tumors and healthy tissues is the baseline oxygen levels, where solid tumors are frequently found to be much less oxygenated than normal tissue. In one of the original FLASH effect theories,²⁷ it was proposed that fast oxygen depletion at UHDRs is able to induce radioprotection of normal tissues by diminishing oxygen fixation of DNA damage,⁶ whereas in tumors, which are already devoid of oxygen, further decrease in oxygenation due to FLASH is not so impactful. Later experimental studies reduced the plausibility of this theory by showing that the amount of oxygen depleted by a clinically relevant dose of radiation, delivered as FLASH, is insufficient to induce sufficient radioprotection.^{12–16,18,19,28} Nonetheless, it was consistently observed in vitro that oxygen depletion caused by FLASH at normoxic conditions (30–60 mmHg) is lower than that caused by an isodose delivered at a conventional dose rate. Both for FLASH and conventional dose rates, g_O2 values decline toward 0 starting around 20 mmHg.^13,14,16,19

On the basis of these findings, an alternative hypothesis about the role of oxygen in the FLASH effect was formulated,¹³ which considers the similarity between the shapes of the oxygen dependencies of g_O2 values and the classic oxygen enhancement ratio curve and assumes that the amount of oxygen consumed due to irradiation is correlated with the inflicted tissue damage. In normoxic conditions, the g_O2 values for conventional dose rates are expected to be higher than those for FLASH, indicating more oxygen consumption and, consequently, more damage to normal tissues by conventional irradiation. However, as baseline pO₂ approaches hypoxia, the g_O2 curves for FLASH and conventional dose rates converge, implying that the amount of oxygen consumed—and thus the tissue damage inflicted—by both dose rates becomes nearly identical. This theory was based entirely on in vitro measurements, whereas the experimental demonstration of the relationship between g_O2 values and baseline tissue pO₂ in vivo had been missing. This work fills this gap by showing that oxygen consumption due to FLASH in vivo has the same dependence on baseline pO₂ as in model solutions in vitro, adding an important piece of evidence in relation to the above hypothesis.

Another important finding of this work is that in tissue in vivo, changes in the frequency of delivery of individual electron pulses (PRF) have no effect on the amount of oxygen consumed, which contrasts with our previous results for model solutions.¹⁶ In this study, 2 PRFs were employed, 120 and 10 Hz, corresponding to the interpulse separation times, Δt, of ~8.3 and ~100 ms, respectively. Previously, it was found that several electron pulses delivered to solutions of albumin in vitro at a lower PRF (60 Hz, Δt = 16.7 ms) resulted in ~15% higher g_O2 values than the same pulses delivered at a higher PRF (360 Hz, Δt = 2.8 ms).¹⁶ Here, we show that unlike in vitro, in tissue in vivo apparent g_O2 values are not measurably dependent on Δt (or PRF), at least for the range of PRFs tested. However, it should be noted that the time resolution of oxygen measurements in this study (0.2 seconds) is significantly lower than the time scale of the relevant radiation-chemical reactions, which occur on the order of microseconds-to-milliseconds. Although this limitation prevents direct observation of rapid oxygen consumption events in the early stages of radiation-chemical reactions, our results demonstrate the impact of initial oxygen tension on the millisecond scale, which remains relevant for assessing net oxygen consumption in relation to the FLASH effect. In addition, fast (sub-millisecond) oxygen measurements performed previously in vitro and in vivo^13,14 showed no evidence of rebound in oxygenation after application of radiation on the timescale faster than 200 ms, which was the time step used in the present measurements.

Recently, El Khatib et al¹⁵ showed that, in the case of protons, g_O2 values measured in the samples of cell pellets, where the probe was located directly in the cell cytoplasm, were nearly equal for conventional and FLASH dose rates. At the same time, in model solutions, either containing albumin¹³ or comprising cytoplasm-mimicking buffers,^14,15 g_O2 values for conventional dose rates were considerably higher than for UHDRs, paralleling the results of in vitro measurements.¹⁶ It was suggested that the difference between the dose rate effects on oxygen consumption in model solutions versus cellular cytoplasm could be associated with the presence of enzymes, such as superoxide dismutase and catalase, which are abundant in the cytoplasm and could affect the net oxygen balance.¹⁵ In this regard, the lack of dependence of g_O2 values on PRF in vivo, observed in this study, could also be caused by enzymatic reactions, and as such it could contain information about the mechanistic aspects of the underlying chemistry initiated by electron pulses at different PRFs. In tissue, the majority of oxygen consumption due to radiation occurs inside cells, because intracellular volume constitutes the dominant fraction of the bulk tissue volume.²⁹ Although in our present experiments the phosphorescent probe was located in the extracellular space, effectively it reported on intracellular oxygen as well due to anticipated rapid equilibration between intracellular and extracellular compartments. Activity of superoxide dismutase and catalase in intact tissues is at least as high as that in isolated cells, and the net oxygen dynamics, which was shown to be different at different PRFs in vitro,¹⁶ was likely influenced by these enzymes and hence reflected in our in vivo measurements.

It is important to note that tumors often have higher baseline levels of oxidative stress compared with normal tissues,³⁰ as well as altered antioxidant defense mechanisms.³¹ Future studies should examine how these intrinsic differences in redox status between tumors and normal tissues may impact oxygen consumption dynamics and potentially influence the differential effects of FLASH irradiation.

To test the above propositions, systematic studies of enzymatic effects on g_O2 values and their dependence on baseline pO₂ in different media types are required. Future work should establish the relevance of oxygen-related phenomena to the essence of the FLASH effect by examining whether there exists a correlation between oxygen consumption at different tissue oxygenations and tissue damage/sparing. Taken together, such studies should improve understanding of the FLASH effect and potentially lead to its optimal application in cancer radiotherapy.

Conclusions

In this in vivo study, radiation-induced oxygen consumption was investigated across a wide range of possible skin tissue oxygenation levels using real-time oximetry techniques. Findings reveal a distinct relationship between tissue oxygenation and radiation-induced oxygen consumption, with consumption rates reaching a plateau of 0.26 ± 0.01 mmHg/Gy above 20 mmHg with no significant dose per pulse or PRF dependence at UHDRs. Ultimately, these observations underscore significant radiation-chemical effect differences based on tissue oxygen tension, with potential implications for the FLASH effect—or lack thereof—in hypoxic tissue.

Data Sharing Statement:

Research data will be shared upon reasonable request to the corresponding author.