In recent years, FFF beams have been increasingly implemented into clinical practice and have shown decisive advantages over flattened beams, especially in SBRT. As a result, several treatment planning studies already investigated the dosimetric effects of FFF beams for SBRT under VMAT conditions in clinical practice. In contrast to the majority of studies performed to date, we used energy-matched FFF beams and a wide range of target volume sizes, comparing four treatment techniques of 6‑ and 10-MV beams with and without FF under VMAT conditions.

Regarding the BOT, a mean time saving of 71.7% and 83.7% in the case of lung lesions and a mean time saving of 63% and 78.4% in the case of liver lesions were demonstrated for 6 MV-FFF and 10 MV-FFF, respectively, relative to the same energy with FF in each case. Thereby, the increasing risk of interplay effects with shortening BOT was countered by the use of multiple arcs and ≥ 2 fractions [37]. Thus, intrafractional movements of the patient or of the tumor can be minimized and techniques such as ABC can be performed in a more patient-friendly manner [13,14,15]. Our findings are consistent with previous studies demonstrating > 50% of BOT reduction for FFF beams, which is largely due to the higher dose rates that can be achieved with FFF beams [17, 22, 23, 26, 38]. Despite the nominally higher maximum dose rate of 10 MV-FFF, we observed only 8% and 11.3% shorter BOT compared to 6 MV-FFF for lung and liver lesions, respectively. Similar results were also obtained by Hrbacek et al. [23], who showed an average 13.3% shorter BOT for 10 MV-FFF compared to 6 MV-FFF using the Varian Truebeam linac, which has maximum dose rates of 1400 MU/min and 2400 MU/min for 6 MV-FFF and 10 MV-FFF, respectively.

When considering dose rate efficiency, a median of 95.4%, 92.8%, 99.9%, and 81.5% for the lung lesions and a median of 94.2%, 89.8%, 99.8%, and 81.5% for the liver lesions of the maximum possible dose rates of 6 MV, 6 MV-FFF, 10 MV, and 10 MV-FFF were achieved, respectively. This matches the findings by Hansen et al. [22], who also used a Versa HD in their planning study for SBRT lung treatments and achieved similar dose rate efficiencies for 6 MV (97.3%) and 6 MV-FFF (89.2%) with dual arc. Consequently, the full potential of the higher dose rates, particularly at 10 MV-FFF, could not be realized, as MLC and gantry speed constraints were limiting factors [25].

Total MUs in lung lesions were significantly lower for FFF beams than for flattened beams of the same energy level, mainly due to the large number of small-volume PTVs. We were able to show that FFF beams in lung lesions up to a PTV of 30.3 ccm required fewer MUs than flattened beams of the same energy. Total MUs in liver lesions were not significantly different for FFF beams than for flattened beams of the same energy level, due to the very heterogeneous size distribution of PTVs. We were able to show a cut-off value of 40.6 ccm comparing 6 MV-FFF versus 6 MV. Comparable results could be shown for 10 MV versus 10 MV-FFF depending on the PTV size.

Previous IMRT planning studies of EBRT of prostate, head and neck, brain, and lung tumors already observed lower MUs for FFF versus flattened beams [39, 40]. However, these results were mainly due to the fact that the linacs were not recalibrated with respect to MUs after FF removal, which is the case for our linac. Hansen et al. [22] also demonstrated a reduction in MUs for energy-matched FFF beams compared to flattened beams in lung treatments with a median PTV of 39.4 ccm (range: 10.8–87.5 ccm). In studies in which energy-matched FFF beams were not used, either no significant differences or significantly more MUs were observed for FFF beams (non-energy-matched). Referring to two studies with a PTV mean of 13.3 ccm (maximum size: 41.1 ccm) and a PTV median of 13.5 ccm (maximum size: 31.3 ccm) investigating mainly small peripheral lung lesions, no significant differences in total MUs between FFF and flattened beams could be demonstrated [28, 31]. Planning studies with a PTV median of 34 ccm (maximum size: 144.5 ccm) and 59.5 ccm (maximum size: 372.43 ccm) examining NSCLCs observed significantly more MUs with FFF versus flattened beams [23, 30]. Another study evaluating liver lesions with a PTV median of 164 ccm (maximum size: 435.1 ccm) showed significantly more MUs with FFF beams compared to flattened beams as well [38].

Liu et al. [30], using the smallest PTV size of 19.01 ccm, did not find a significant correlation between MUs and PTV. Indeed, Vieillevigne et al. [29] as well as Hrbacek et al. [23] showed a systematic increase in MU ratio (FFF/FF) with increasing PTV size.

Considering the graphical presentation of the output factor as a function of field size, as shown by Paynter et al. [32], a comparable phenomenon can be observed. At small field sizes, flattened beams require more MUs per Gy. The beam profile of flattened and FFF beams hardly differs at small field sizes, and FFF beams exhibit lower scattering. As the field size increases, flattened beams require fewer MUs per Gy compared to FFF beams. With increasing field size, the different beam profiles between flattened and FFF beams also become relevant, such that for FFF beams more MUs are needed to achieve the same dose coverage due to the conical beam profile. Apart from this effect being more pronounced in energy-matched than for non-energy-matched beams [32], other vendor- (e.g., MLC design, TPS, segmentation algorithm [41]) and patient-specific differences (e.g., PTV size/geometry, PTV distance from OAR) should also be noted.

Comparable target volume coverage was achieved for all plans. Nevertheless, for the lung lesions we showed a significant, but low improvement in GTV coverage for FFF beams and a significant better PTV coverage for 10 MV-FFF compared to 10 MV. For liver lesions, we found slightly superior GTV coverage for 10 MV-FFF compared to 10 MV and 6 MV-FFF, and better PTV coverage for FFF beams.

This is in line with the findings of Tambe et al. [24] and Liu et al. [30], who also demonstrated a minimally superior PTV coverage (< 1%) for FFF beams. By contrast, other studies could not detect any differences between flattened and FFF beams with respect to target volume coverage [22, 23, 26].

With regard to PTV conformity, no significant differences were found between FFF and flattened beams in lung and liver lesions, which is consistent with other results [22, 26, 29, 31, 38], whereas Hrbacek et al. [23] demonstrated significantly better conformity for FFF beams in their investigations. It should be noted that different CIs were used in each study, making the results comparable only to a limited extent.

Due to the very small target volumes in our patients with lung lesions, it was difficult to achieve CIs smaller than 1.2, which Vieillevigne et al. [29] also described. For the larger liver lesions, most CI values were < 1.2.

With respect to dose fall-off, our data suggest that the largest differences occur in the medium to lower dose range. Apart from the 6 MV-FFF GIP&L being preferable over that for 10 MV-FFF in lung lesions, no significant differences for the GIs between the beam selections were found in lung and liver lesions. Barbiero et al. ([31]; GI = PIV50%PIV/PIV95%PIV) and Zhang et al. ([25]; GI = GIP&L) also came to the same conclusion when evaluating their GIs. By contrast, other studies showed significantly lower GIs (GIRTOG [22,23,24, 28], GIP&L [29]) for FFF versus flattened beams.

Our results for lung lesions indicated a significantly steeper dose fall-off from the prescribed dose to shell 2 and 3 for FFF beams. In addition, the integral dose (D50% and Dmean) was significantly lower for FFF versus flattened beams of the same energy in lungs as well as liver lesions, which matches the observations made by Hansen et al. [22].

These findings can be explained by the reduction of scattering, leaf transmission, and head leakage, which lead to lower out-of-field dose for FFF compared to flattened beams [18,19,20]. Furthermore, when comparing 6 MV-FFF and 10 MV-FFF, we found a lower dose exposure of healthy tissue for 6 MV-FFF.

The evaluation of the OAR of the lung cohort was performed in two subgroups that, to our knowledge, no study has investigated before. In the first subgroup, with centrally or superiorly located lung lesions, a tendency of a lower dose exposure to the contralateral lung for 6 MV-FFF as well as significant results for 10 MV-FFF relative to the irradiation technique of the same energy level with FF were found. This was also the case in the second subgroup, the peripheral lung lesions, for 6 MV-FFF versus 6 MV.

Hansen et al. [22] and Tambe et al. [24] reported comparable results, but also showed significant dose savings for 6 MV-FFF versus 6 MV in the tumorous lung, which cannot be confirmed by our data so far. In addition, for the first subgroup 10 MV-FFF offered superior dose savings of the total lung and myelon compared to 10 MV. Consistent with the results of Lu et al. [27] and Tambe et al. [24], we found a significantly higher dose exposure to the lung for 10 MV-FFF versus 6 MV-FFF. In peripheral lung lesions, comparable results were found with respect to the Dmean of the total as well as tumorous lung.

Regarding liver lesions, a significantly lower dose exposure to the liver (D700ccm) was demonstrated for FFF beams compared to flattened beams of the same energy level. This is consistent with the results of Vieillevigne et al. [29]. Further dose savings in OAR (i.e., myelon, kidney left) of liver lesions were achieved for 6 MV-FFF compared to 6 MV. Overall, significant differences were observed especially in OAR in the lower dose range, as was already shown by our findings regarding dose fall-off in healthy tissue. Nonetheless, even a small dose reduction is worthwhile and lowers the risk of secondary cancers [42].

Compared to the other beam selections, 10 MV-FFF showed the lowest dose exposure of the skin. With 6 MV-FFF, on the other hand, we found the highest average skin dose. Other studies also indicate superior skin protection for 10 MV-FFF compared to 6 MV-FFF and 6 MV [15, 24, 43].

Due to the lack of beam hardening effect and the resulting softer photon spectrum with FFF beams, a higher skin dose can be expected. This is partially offset by the reduction in head scatter and leaf leakage due to the absence of the FF, which contributes to a lower surface dose [44].

Since energy-matched FFF beams were used in our study and consequently have a higher energy than unmatched FFF beams, we found a skin dose saving even for 10 MV-FFF versus 10 MV. On the other hand, for 6 MV-FFF versus 6 MV, the skin dose was mostly higher. Apparently, for energy-matched 6 MV-FFF the missing beam hardening effect still outweighs the favorable characteristics such as reduction in head scatter, head leakage, and leaf transmission of the FFF beams.

Limitations of our study are mainly due to the retrospective design and the limited patient population included in our liver cohort. Nevertheless, our investigations show the influence of vendor-specific (e.g., linac, MLC, gantry speed, TPS, dose calculation algorithm) as well as patient-specific (e.g., PTV size/geometry, distance to OAR, irradiation scheme) differences in the use of FFF beams in SBRT. This is of great interest for personalized medicine.

In this context, the multicenter planning studies of the DGMP Working Group have shown that these dosimetric differences depend more on techniques (IMAT, 3D-CRT, SF-IMRT, robotic radiosurgery) than on the planners [45]. The multiparameter dose prescription with normalization to GTV or ITV mean or median dose, pointed out by the authors, can harmonize the differences between different energies, flattened and unflattened, and matched and unmatched FFF beams, leading to further standardization of SBRT [45, 46].