FLASH Sparing Effect in Radiotherapy

• FLASH sparing effect is the reduction of normal-tissue toxicity when ultra-high dose rate irradiation is delivered, maintaining or sometimes improving tumor control.
• Empirical studies identify key parameters such as total dose, delivery time, and radical recombination as critical in triggering the sparing phenomenon.
• Mechanistic frameworks centered on oxygen depletion and radical-density interactions offer practical insights for optimizing clinical translation.

The FLASH sparing effect denotes the reduction in normal-tissue toxicity observed when irradiation is delivered at ultra-high dose rate, typically at mean dose rates of at least $40\,\mathrm{Gy/s}$, while tumor control is maintained or sometimes improved relative to conventional dose-rate irradiation. Across the current literature, the effect is operationally tied not only to mean dose rate but also to total dose, instantaneous dose rate, pulse structure, delivery time, tissue oxygenation, and radical chemistry. It has been reported with electrons, protons, photons, and heavy ions, but its trigger conditions and mechanistic basis remain context-dependent rather than universal (Ma et al., 2024, Zhang et al., 2024).

1. Empirical definition and observed scope

In contemporary radiobiology, FLASH radiotherapy is generally defined as irradiation delivered at ultra-high dose rate in short bursts that complete a clinically relevant dose within tens to hundreds of milliseconds. The associated sparing effect is the phenomenon of marked normal-tissue protection with maintained—or sometimes improved—tumor control compared with conventional dose-rate irradiation. Preclinical reports encompass brain, lung, skin and soft tissues, and the gastrointestinal tract, including preservation of memory and cognitive function, reduced fibrosis, reduction in dermatitis and contracture, and intestinal crypt sparing. Multiple modalities have been used, including electrons, protons with Spread-Out Bragg Peak delivery, synchrotron and megavoltage X-rays, and heavy ions such as carbon and helium (Ma et al., 2024).

The empirical scope is broad but not uniform. Not all studies observe sparing, and negative results have appeared with electrons and protons under certain geometries, targets, or dosimetry. A consistent conclusion from the review literature is that average dose rate alone is not sufficient: total dose, instantaneous dose rate, dose-per-pulse, pulse number and spacing, irradiation time, beam depth and energy, and tissue context all affect whether a measurable sparing phenotype appears (Ma et al., 2024).

2. Quantitative trigger conditions and the temporal reaction window

A controlled whole-abdomen mouse study using a $6\,\mathrm{MeV}$ electron beam provides one of the clearest threshold characterizations of the FLASH sparing effect at an acute intestinal toxicity endpoint. In female C57BL/6 mice, with sparing defined by a composite histological score at 72 hours, no discernible FLASH sparing was detected at $6$, $9$, or $12\,\mathrm{Gy}$ when the mean dose rate was fixed at $900\,\mathrm{Gy/s}$; a significant sparing effect emerged only at $15\,\mathrm{Gy}$. At fixed total dose $D=15\,\mathrm{Gy}$, sparing was observed only when the mean dose rate reached at least $100\,\mathrm{Gy/s}$, corresponding to a delivery time $T = D/\dot{D} \approx 150\,\mathrm{ms}$. The magnitude of sparing increased with mean dose rate and plateaued once the mean dose rate exceeded $6\,\mathrm{MeV}$0, corresponding to $6\,\mathrm{MeV}$1. Under these same conditions, changing dose per pulse from $6\,\mathrm{MeV}$2 to $6\,\mathrm{MeV}$3 at fixed total dose and mean dose rate produced no significant difference in histological score, indicating that pulse structure was not the decisive variable in that regime (Zhang et al., 2024).

The same study linked these thresholds to a characteristic reaction window. The recombination fraction of peroxyl radicals showed an S-shaped dependence on delivery time, with a midpoint of change around $6\,\mathrm{MeV}$4 and saturation for $6\,\mathrm{MeV}$5. This temporal dependence matched the observed biological thresholds: no sparing at $6\,\mathrm{MeV}$6 ($6\,\mathrm{MeV}$7), onset at $6\,\mathrm{MeV}$8, and saturation by at least $6\,\mathrm{MeV}$9 (Zhang et al., 2024).

Subsequent work in the same intestinal model further sharpened the notion of a dose threshold. At a total dose of $6$0, combined FLASH+CONV schedules retained sparing when the FLASH component was at least $6$1, whereas $6$2 of FLASH plus $6$3 of conventional irradiation lost sparing and resembled conventional $6$4. At $6$5 total, no sparing was observed regardless of FLASH/CONV partition. This suggests that the relevant condition is not merely the presence of an ultra-high-dose-rate component, but the delivery of an adequate total radical-generating dose within the effective reaction window (Zhang et al., 28 Aug 2025).

3. Physicochemical frameworks: radical recombination, oxygen depletion, and interspur interactions

Two physicochemical frameworks dominate current mechanistic discussion. One is radiolytic oxygen depletion, formalized as

$6$6

with reported depletion yields in the range $6$7–$6$8. For $6$9 and $9$0, the expected drop is $9$1, which is only a transient $9$2 reduction from a baseline $9$3. This is generally regarded as too small to explain large sparing by oxygen effects alone in well-oxygenated tissues, although the effect is more consequential in already hypoxic contexts (Ma et al., 2024).

The second framework is radical-density-driven recombination, especially involving peroxyl radicals. In the intestinal mouse model, a representative kinetic equation was written as

$9$4

At conventional dose rate ($9$5), the simulated fraction of $9$6 recombination was essentially zero across doses, whereas under FLASH ($9$7) the recombination fraction increased approximately linearly with total dose. Correspondingly, protein hydroperoxide ($9$8) at FLASH $9$9 nearly matched $12\,\mathrm{Gy}$0 in conventional $12\,\mathrm{Gy}$1, paralleling the histological finding that FLASH $12\,\mathrm{Gy}$2 produced damage comparable to conventional $12\,\mathrm{Gy}$3 (Zhang et al., 2024).

Analytical work on interspur interactions reaches a related conclusion from a different direction. By modeling the overlap of radiolytic spurs from different tracks, that framework predicts a minimum critical dose and dose rate for significant interspur interaction, and it predicts that interspur interactions correlate negatively with beam linear energy transfer at fixed dose. This supports the broader proposition that FLASH depends on high transient radical density and early-time intermolecular competition, not solely on average dose rate (Baikalov et al., 2022).

4. Molecular and cellular correlates

At the molecular level, a DNA-based FLASH-like sparing effect has now been demonstrated under physiologically realistic oxygen, salt, and pH conditions. In plasmid pUC19 irradiated with $12\,\mathrm{Gy}$4 electrons in $12\,\mathrm{Gy}$5PBS at pH 7.4, under physoxia of $12\,\mathrm{Gy}$6–$12\,\mathrm{Gy}$7 oxygen, measurable sparing emerged above approximately $12\,\mathrm{Gy}$8, and the most conspicuous separation between low dose rate and ultra-high dose rate occurred above approximately $12\,\mathrm{Gy}$9–$900\,\mathrm{Gy/s}$0. The loss of supercoiled plasmids shifted from $900\,\mathrm{Gy/s}$1–$900\,\mathrm{Gy/s}$2 under low dose rate to $900\,\mathrm{Gy/s}$3 under ultra-high dose rate. Under ambient oxygen of $900\,\mathrm{Gy/s}$4–$900\,\mathrm{Gy/s}$5, this behavior was absent. The accompanying TOPAS-nBio analysis attributed the effect to reduced $900\,\mathrm{Gy/s}$6 availability, enhanced recombination under dense spur overlap, and reduced transient $900\,\mathrm{Gy/s}$7, which in turn decreased $900\,\mathrm{Gy/s}$8-elimination-driven strand-break conversion (Hahn et al., 17 Oct 2025).

At the cellular and tissue level, multiple downstream correlates have been reported. FLASH irradiation has been associated with reduced cytoplasmic double-stranded DNA and attenuated cGAS–STING activation in normal intestinal crypts, reduced $900\,\mathrm{Gy/s}$9H2AX formation at higher doses in some proton studies, preservation of mitochondrial function in normal cells, cytochrome $15\,\mathrm{Gy}$0 release with suppression of mtDNA-driven IFN-$15\,\mathrm{Gy}$1 signaling, preservation of vascular integrity, reduced microglial activation, and lower TGF-$15\,\mathrm{Gy}$2 in normal tissues. These observations have been interpreted as a multi-scale sequence in which altered early radical chemistry propagates into reduced persistent ROS, altered DNA damage processing, and attenuated inflammatory and fibrotic responses (Ma et al., 2024).

Redox-regulatory signaling has been added to this picture in the intestine. FLASH-RT enhanced peroxyl radical recombination, reduced ROS and malondialdehyde, increased p-NRF2/NRF2, HO-1, Hmox1, Gsta1, GSH/GSSG, and catalase activity, and suppressed p-ERK and cleaved caspase-3. NRF2 knockdown abolished the differential survival between conventional and FLASH irradiation in FHS74Int cells, placing antioxidant defense downstream of the early radical chemistry and upstream of the tissue-level sparing phenotype (Zhang et al., 28 Aug 2025).

5. Antioxidant perturbation and causal inference

Antioxidant intervention has become one of the strongest mechanistic probes of the FLASH sparing effect. In the mouse intestine, amifostine administered at $15\,\mathrm{Gy}$3 intraperitoneally 15–30 minutes before irradiation markedly reduced conventional-rate injury to a level comparable to FLASH without amifostine, while the FLASH group showed little further change with amifostine. Histology showed preserved crypt structure and reduced inflammatory infiltration, and the histological scores showed no significant difference between FLASH with and without amifostine. The interpretation given was that scavenging radicals and intercepting propagation competes with the recombination pathway: if radical damage is already curtailed by FLASH-induced recombination, adding an antioxidant confers limited additional benefit; if it is not, antioxidants substantially lower damage (Zhang et al., 2024).

Parallel results were obtained with both amifostine and N-acetylcysteine in the later intestinal study. Amifostine improved both conventional and FLASH groups at $15\,\mathrm{Gy}$4, but conventional plus amifostine reached FLASH-like histology and Ki-67 levels, while NAC raised conventional cell survival to FLASH levels without further benefit in the FLASH condition. Combined with the NRF2 knockdown result, these experiments support a free-radical-mediated mechanism in which FLASH alters the balance between radical generation, radical recombination, and antioxidant interception (Zhang et al., 28 Aug 2025).

These data do not exclude oxygen-dependent contributions. They do, however, show that in at least one well-defined intestinal model, the observed thresholds, delivery-time dependence, and loss of FLASH–CONV difference after antioxidant administration can be captured without invoking oxygen depletion as the primary explanatory variable. This suggests that free-radical recombination is sufficient in that setting, though not necessarily exclusive (Zhang et al., 2024).

6. Delivery metrics, planning formalisms, and translational implications

A recurring translational point is that “FLASH conditions” cannot be reduced to a single reported average dose rate. The review literature emphasizes rigorous characterization of average and instantaneous dose rate, pulse number, pulse width, spacing, total dose, delivery time, beam energy, irradiation depth, region, and tumor status. Dosimetry at ultra-high dose rate remains technically difficult because detector saturation and ion chamber recombination can bias measurements, and oxygen measurements require millisecond-scale resolution to capture transient depletion (Ma et al., 2024).

The distinction between instantaneous and average dose rate has become especially salient in proton studies. Laser-driven protons delivered $15\,\mathrm{Gy}$5 in $15\,\mathrm{Gy}$6 per bunch, yielding an instantaneous dose rate of approximately $15\,\mathrm{Gy}$7, while the average dose rate was only $15\,\mathrm{Gy}$8 because bunches were separated by $15\,\mathrm{Gy}$9. In a BALB/c mouse ear model, this delivery reduced tissue swelling relative to conventional X-rays, with $D=15\,\mathrm{Gy}$0 laser-driven protons showing significant sparing versus $D=15\,\mathrm{Gy}$1 X-rays. This finding does not redefine FLASH, but it indicates that ultrashort, ultra-high instantaneous dose-rate bunches may have biological significance even when the average dose rate is below the conventional $D=15\,\mathrm{Gy}$2 criterion (Obst-Huebl et al., 24 Feb 2026).

Treatment-planning studies have translated the sparing concept into organ-at-risk models. For proton pencil-beam scanning SBRT in spine metastasis, one study assumed that spinal-cord protection would occur when both dose rate and per-fraction dose thresholds were met, using $D=15\,\mathrm{Gy}$3 and $D=15\,\mathrm{Gy}$4 per fraction as operational criteria. The planning formalism defined a FLASH modifying factor,

$D=15\,\mathrm{Gy}$5

and concluded that values of approximately $D=15\,\mathrm{Gy}$6 to $D=15\,\mathrm{Gy}$7 would be required to render the spinal-cord DVH compliant in the modeled case (Wuyckens et al., 2024). A separate proton SFRT-FLASH planning framework enforced per-field thresholds of $D=15\,\mathrm{Gy}$8 and $D=15\,\mathrm{Gy}$9, achieving FLASH coverage of approximately $100\,\mathrm{Gy/s}$0–$100\,\mathrm{Gy/s}$1 in a $100\,\mathrm{Gy/s}$2 ring around the CTV while preserving peak-to-valley dose ratios of approximately $100\,\mathrm{Gy/s}$3–$100\,\mathrm{Gy/s}$4 at shallow-to-intermediate depths (Zhang et al., 9 May 2025). More generally, if the underlying mechanism is local, organ architecture matters: an NTCP modeling study found that for a fixed conventional NTCP of $100\,\mathrm{Gy/s}$5, the corresponding FLASH NTCP ranged from $100\,\mathrm{Gy/s}$6 for $100\,\mathrm{Gy/s}$7 to $100\,\mathrm{Gy/s}$8 for $100\,\mathrm{Gy/s}$9, implying greater sparing in serial than in parallel organs (Pardo-Montero, 28 Apr 2026).

7. Controversies, competing explanations, and open problems

The principal controversy is not whether a delivery-dependent sparing phenomenon exists, but which mechanism is sufficient under which conditions. ROD-based models consistently predict that radioprotection should be strongest in already hypoxic tissue, that single-pulse delivery or multi-pulse delivery with short pulse intervals should maximize the effect, and that low-LET beams should favor ROD-mediated sparing (Zhu et al., 2021, Pratx et al., 2019). A related spatial-heterogeneity model shows how even mean oxygen drops of less than $T = D/\dot{D} \approx 150\,\mathrm{ms}$0 can produce large survival changes if rare nearly hypoxic microregions dominate response, especially in irregular capillary networks (Taylor et al., 2022). Yet tumor-control modeling based on ROD alone predicts lower TCP under FLASH than under conventional irradiation at the same physical dose, particularly in hypoxic tumors and low $T = D/\dot{D} \approx 150\,\mathrm{ms}$1 settings, indicating that ROD alone does not straightforwardly explain iso-effective tumor control (González-Crespo et al., 2023).

Another open problem concerns simplified aqueous chemistry readouts, especially hydrogen peroxide yields. Older pulse-radiolysis experiments and several Monte Carlo frameworks generally predict that $T = D/\dot{D} \approx 150\,\mathrm{ms}$2 yields increase with dose rate in oxygenated pure water once total dose exceeds approximately $T = D/\dot{D} \approx 150\,\mathrm{ms}$3, whereas several recent experiments in plastic containers and protein-containing systems report lower $T = D/\dot{D} \approx 150\,\mathrm{ms}$4 at ultra-high dose rate. The current assessment is that experimental confounders such as container-derived scavengers, oxygen release, assay pH sensitivity, and incomplete pulse-structure reporting may contribute materially to the disagreement (Hahn, 30 Oct 2025).

More speculative physical theories also remain in circulation. One recent proposal attributes the normal–tumor differential to morphology-dependent charge recombination, arguing that ordered normal tissues can cross a threshold for classical electron–hole liquid formation under ultra-high dose rate, thereby suppressing free-radical generation, whereas disordered tumor tissues do not (Shvydka et al., 16 Mar 2026). This hypothesis supplies explicit dose and dose-rate threshold formulas, but it remains conceptually distinct from the radical-recombination and oxygen-depletion frameworks that currently dominate experimental work.

Taken together, the literature supports a robust but conditional definition of the FLASH sparing effect: normal-tissue protection emerges when a sufficiently large dose is delivered within a short physicochemical reaction window, and the magnitude of the effect depends on how beam delivery reshapes radical competition, oxygen chemistry, and downstream stress signaling in a tissue-specific microenvironment. The unresolved problem is not the existence of thresholds, but the quantitative mapping from thresholds to mechanism across modalities, tissues, and clinical geometries.