Shining a FLASHlight on Ultrahigh Dose-Rate Radiation and Possible Late Toxicity

In this issue of Clinical Cancer Research, Rohrer Bley and colleagues (1) describe late toxicity from FLASH radiotherapy (FLASH-RT) in cats with squamous cell carcinoma (SCC) and mini-pigs. “FLASH” refers to radiotherapy delivered at ultrahigh dose rates (≥40 Gy/s), so that treatment occurs within milliseconds rather than minutes as with conventional dose rate (Conv-RT). It was coined by this same group in a seminal article in 2014 showing that FLASH-RT of mice using electron beams protected normal lung from fibrosis compared with radiotherapy Conv-RT (2). In the past 8 years, this group and others have observed reduced toxicity in diverse tissues in animals, including the brain, intestine, and skin using electron (3–5), proton (6–10) and X-ray irradiation (11).

Experimental data suggest a 20%–40% reduction in normal tissue toxicity using FLASH-RT versus Conv-RT with equipotence in tumor growth control (12). Hence, there is tremendous interest in clinical translation; however, there are multiple hurdles to overcome. First, there are technical challenges in delivering FLASH-RT to the larger irradiated volumes in human patients and then developing quality assurance technology to ensure that the correct dose rates/doses have been delivered to the correct prescribed volume (13). Second, most preclinical studies have been performed in mice using small fields and a single dose, which is uncommon in human patients, so the generalizability of these results is untested. Hence, before transitioning from mouse to human, it is desirable to demonstrate similar sparing with FLASH-RT in larger, non-human vertebrates. Hence, the current article (1) is an important step in this journey.

Previously, Vozenin and colleagues (14) showed results from a phase I FLASH dose-escalation study (25–41 Gy) for cats with SCC of the nasal planum. Three subjects experienced no acute toxicity, and three exhibited moderate/mild transient mucositis. All had depilation, but no other late toxicity was observed (median follow 18 months). Mini-pigs given 24–34 Gy to a 2.6-cm diameter on the skin of the back exhibited depilation. FLASH-RT preserved hair follicles whereas Conv-RT permanently destroyed follicles with no hair regrowth for >6 months and was associated with severe late skin fibronecrosis.

In 2019, this group started a phase III clinical trial, whose results are now reported (1). Cats with nasal planum SCC were randomized to 4.8 Gy × 10 treatments (Conv-RT; standard of care) or a single 31 Gy FLASH-RT treatment. Recruitment was stopped after treatment of 16 cats (as per predetermined criteria), because the first FLASH-treated cat patient developed bone necrosis (grade 3 late toxicity). Local control was excellent with only one cat in each group developing local recurrence (median follow-up 19.7 months). Acute/subacute toxicity ranged from mild–moderate in both groups. Unexpectedly, three of the 7 cats in the FLASH arm exhibited late high-grade toxicity, but none of the Conv-RT cats. All three FLASH-RT–treated cats developed bone necrosis at 12.5–15.1 months (two after developing grade 3 mucosal breakdown and one after severe atrophy/necrosis of the skin), leading to premature cessation of the trial.

The current study (1) also includes results of FLASH-RT on mini-pigs to larger fields than used previously (14). FLASH-RT of 3.5×4.5 cm fields resulted in late skin lesions evolving from erythema and ulceration to permanent hyperkeratosis and skin contracture (6–8 months post-RT). Furthermore, pigs treated to an 8×8-cm field developed severe skin reaction with telangiectasia at 5 months post-RT, evolving into progressive epithelial ulceration (6 months) and necrotic scabs (7–9 months). Hence, these results suggest that late toxicity may be more pronounced when larger field sizes are used, which is relevant to humans.

The current study has some limitations: First, due to the nature of electron beam and the high-dose used, a longer duration treatment was used—3 pulses over 20 millisecond versus a single dose over microseconds, which in previous studies led to normal tissue sparing in mice and zebrafish (5, 12, 15). The temporal pulse structure (# pulses, time between pulses, delivery time) of electron FLASH can vary significantly (12, 16) and may be critical for normal tissue sparing, as suggested by a study examining intestinal crypt survival after abdominal electron FLASH-RT in mice (4). Second, 31 Gy x 1 was used for FLASH-RT, but 4.8 Gy × 10 for Conv-RT. A more appropriate comparison would have been to use fractionated regimens for both modalities. The reason for delivering radiotherapy in multiple smaller doses rather than a single large dose is that decades of clinical experience have shown that fractionation can reduce late toxicity. However, this has been done using Conv-RTs, and we do not know how delivering FLASH-RT in multiple fractions will affect late toxicity. It is speculated that there is a minimal threshold below which FLASH-RT normal tissue sparing effects will not be seen, perhaps in the 7–10 Gy range (15, 16). If true, FLASH-RT would not be appropriate for standard definitive fractionation schemes (typically 2 Gy daily doses) or the 3 Gy × 10 regimen commonly used in palliative settings (Fig. 1A). It may, however, be effective in a hypofractionated regimen incorporating higher daily doses, for example, 10 Gy × 3. Notably, data show that fractionated FLASH RT (7 Gy × 2) is equipotent with fractionated Conv-RT in growth delay in an orthotopic mouse glioblastoma model (16). Third, it should be noted that electrons have a much broader scattering profile when they enter tissue compared with photons and protons. The latter also have the added benefit of depositing energy within a confined space with no “exit dose” when delivered as a Bragg Peak, with the potential to limit dose deposition in critical tissues. Finally, it was clearly beyond the scope of this study to offer any mechanistic explanation as how osteonecrosis occurred or how, in theory, FLASH might prevent this, but work has been done to understand bone toxicity secondary to conventional RT (ref. 17; Fig. 1B). Efforts to move FLASH-RT into the clinic need to be accompanied by continuing studies to understand its biological basis (18).

Despite these limitations, the importance of these findings cannot be overemphasized, as human clinical trials are already underway. A phase I trial treating bone metastases with FLASH RT was recently completed (FAST-01 trial; NCT04592887). Although the results have not yet been published, it has been reported that there have been no stops in the trial due to acute toxicities. A notable difference between FAST-01 and the current cat trial (1) is that the former used protons instead of electrons, and the dose was significantly lower (8 Gy vs. 31 Gy). There is also an active dose-escalation trial open using electron FLASH-RT for cutaneous melanoma metastases (NCT04986696).

Clearly, the data presented by Rohrer Bley and colleagues (1), should raise caution and prompt careful consideration of the design of clinical trials. In light of their study, in future human trials in which larger volumes, including bone or other critical structures are treated, it will be critical to carefully consider tumor/normal tissue site, dose-escalation approaches, dose/fractionation and to have sufficiently long follow-up to assess late toxicity. However, their results should not dissuade further development of this promising modality that could result in reduced side effects affecting millions of patients with cancer undergoing radiotherapy treatments worldwide.

C. Koumenis reports grants from IBA outside the submitted work. No potential conflicts of interest were disclosed by the other author.

This work was supported by NCI P01CA257904 grant (to C. Koumenis and A. Maity). IBA provided the engineering support that allowed the FLASH experiments to be performed.