An in vivo -validated dual-mechanism model explains and guides FLASH radiotherapy for normal-tissue sparing

The clinical effectiveness of radiotherapy (RT) is often limited by normal tissue toxicity. FLASH-RT, delivered at ultra-high dose rate (UHDR, >40 Gy/s), can markedly reduce normal tissue injury without compromising tumor control-a phenomenon known as the FLASH effect. Despite considerable interest in FLASH clinical translation, a critical challenge remains, the mechanisms underlying the FLASH effect are not understood. Published studies are often confounded by differences in organ type, endpoints, and beam parameters. This uncertainty impacts the selection of dose rate and dose required to optimize FLASH effect. Normal tissue sparing under FLASH-RT is widely attributed to rapid physicochemical reactions. Two leading hypotheses have emerged: radiolytic oxygen depletion (ROD), in which transient O₂ depletion reduces the oxygen enhancement ratio (OER) and DNA damage; and radical-radical recombination (RRR), in which elevated radical concentrations promote recombination of lipid peroxyl radicals, suppressing lipid peroxidation (LP) and apoptotic signaling. However, neither mechanism alone fully explains experimental observations. We propose that the FLASH sparing effect could arise from the synergistic contributions of ROD and RRR, as both affect cellular damage through distinct pathways. Given the complexity of physicochemical reactions induced by irradiation, computational modeling is essential in elucidating FLASH mechanisms. We developed a physicochemical model that integrates both mechanisms. The OER-weighted dose serves as a surrogate for DNA double strand break (D DSB ), while the dose resulting in LOOH formation quantifies LP-related toxicity (D LP ). The overall tissue toxicity is represented by the Damage Equivalent Dose (DED) as sum of D DSB and D LP . Our results show that DED generates a consistent normal tissue complication probability (NTCP) curve that accurately captures published data on FLASH sparing of acute GI toxicity and late brain toxicity, across a wide range of UHDR and conventional dose rate (CONV) conditions, whereas dose alone fails to do so. Furthermore, the brain toxicity correlates with D LP , supporting an RRR-dominated mechanism, whereas GI toxicity correlates with both D DSB and D LP , indicating synergistic contributions of ROD and RRR. We further present iso-FLASH tissue sparing maps that delineate dose and dose-rate ranges where FLASH sparing is expected, stratified by organ type and endpoint. Finally, we apply this model to investigate the effects of average dose rate and dose per pulse reported in abdominal irradiation studies, providing insights into parameters governing FLASH-mediated tissue sparing. In sum, this physicochemical model provides an unified and important mechanistic framework that advances our understanding of FLASH-RT sparing effects and guides the optimization of dosimetric parameters for clinical translation.