Effect of simultaneous integrated boost concepts on photoneutron and distant out-of-field doses in VMAT for prostate cancer

Recently, moderately hypofractionated intensity modulated radiotherapy (IMRT) regimens were assigned as standard of care in interdisciplinary national and international guidelines [1, 2] for prostate cancer treatments. We are currently faced with an increasing number of simultaneous integrated boost (SIB) concepts being used both in clinical trials and in daily routine (clinicaltrials.gov yielded 85 active or recruiting trials for “simultaneous integrated boost”). Evidence for these approaches is drawn from moderate hypofractionation trials for prostate [3‐8]. For these approaches, an SIB concept is often the method of choice. The SIB increases fractional radiation dose in selected regions of the treated volume. This leads to shortening of overall treatment time and results in steeper dose gradients toward more radiosensitive organs and consequently a higher dose conformity [9]. Furthermore, moderately increased fractional doses potentially result in beneficial fractionation effects in tumors with low alpha/beta values.

High-energy photons may, depending on their energy, generate photoneutrons (PN) when interacting with beam-shaping components in the medical linear accelerator (linac) treatment head. The primary PN-generating structures were identified [10] to be the radiation target, primary collimator, flattening filter, multileaf collimator (MLC), and jaws. The generated PN typically possess kinetic energies between 0.2 and 2.5 MeV. To account for the higher radiobiological effectiveness (RBE) of neutron radiation, the physical dose D in Gray (Gy) is converted into the equivalent dose H in Sievert (Sv) by multiplication of D with a weighting factor, which is between 2.5 and 20.7 (ICRP 103, [11]). Due to the increased RBE, PN have undesired biological and physical effects on surrounding healthy tissue but also on implanted electronic devices such as cardiac pacemakers and cardioverter-defibrillators, insulin pumps, or neural stimulators (AAPM158, [12]).

With respect to radiation protection, it is furthermore relevant to note that matter within the treatment beam leads to formation of scattered radiation that, in combination with radiation transmitted through the beam-shaping devices, leads to distant out-of-field dose (D_OOF) in the patient. The OOF radiation contributes to potentially unmonitored and unaccounted radiation exposure at a distance from the treated volume, which means there is an increased risk for secondary cancers in organs distant to the primary cancer region [13]. This dose remains unmonitored because not all treatment planning systems (TPS) explicitly account for distant scattered radiation [14] and if they do, CT scans typically do not exceed the adjacent body regions closest to the target volume.

In conventional three-dimensional conformal radiation therapy, the flattening filter (FF) generates a uniform photon fluence within the therapeutic beam. Here, up to 40 × 40 cm² uniform photon beams are shaped through position-dependent beam attenuation into a conformal radiation field. In intensity modulated radiation therapy (IMRT), several smaller segments of irregular shaped fields and with differing photon fluences are directed from several angles to superimpose to a uniform dose with high conformity to the treatment volume. Modern IMRT can be applied with continuously moving multileaf collimator (MLC) leaves and jaws during continuously changing gantry positions and varying dose rates (e.g., VMAT [volumetric modulated arc therapy]). While a wedge becomes unnecessary in modulated segmented radiation fields, the FF frequently remains in the beam. Removal of the FF allows for high dose rates (15–25 Gy/min) that are beneficial for irradiation of small and potentially moving volumes with tight margins as present in stereotactic body radiotherapy at the cost of a uniform lateral target fluence profile. Additionally, removal of the FF in IMRT reduces matter in the beam, hence reducing the contamination with PN and also scattered radiation at a distance from the treated volume [13, 15‐17].

All the aforementioned modifications (SIB, FFF) result in varying modulation depths (number of required monitor units [MU] divided by prescribed dose to target) and differing average equivalent square field sizes (A_eq) describing MLC activity. This might influence PN generation and D_OOF. For locations distant to the isocenter, no studies exist that describe the effect of a prostate SIB in combination with FFF-VMAT in view of neutron contamination or scattered radiation. This might be significant especially for implanted electronic devices in this distance from radiation volumes such as cardiac implanted electronic devices (CIEDs; [18, 19]). In parallel to the increasing use for SIB concepts, there is currently an ongoing trend toward hypofractionation in general, and for prostate radiation therapy (RT) in particular. We therefore addressed the question of whether this trend is accompanied by an increase in out-of-field doses originating from either scattered photons or photoneutrons. In this context, we investigated the effect of an SIB concept in flattened and flattening-filter free VMAT with 6 and 10 MV for prostate cancer RT on distant neutron- and out-of-field dose in VMAT plans of comparable dose distributions and comparable therapeutic efficacy. The results of this study might influence the decision process in whether RT for prostate carcinoma should be moderately hypofractionated, especially in CIED-bearing individuals, or not.

The resulting neutron-equivalent doses H_PN per delivered treatment fraction as a function of either MUs per treatment plan (Fig. 2a) or weighted equivalent square field size (Fig. 2b) are presented in Fig. 2. In analogy, the resulting out-of-field doses D_OOF are presented in Fig. 2c and d. In all panels, the eight different treatment scenarios with all 30 treatment sequences are displayed. Mean values are displayed in Fig. 2 while the standard deviations as percent of the corresponding mean value were σ(H_PN) = 0.80% (between 0.44 and 1.47% for all groups) and σ(D_OOF) = 0.54% (between 0.26 and 0.91% for all groups). Figure 3 shows violin and box plots of the 30 treatment plans for each of the eight treatment approaches. Neutron-equivalent doses per treatment fraction (panel a) and per MU (panel b) as well as out-of-field doses per treatment fraction (panel c) and per MU (panel d) are displayed. Boxplots indicate one interquartile range (IQR).

The prescription dose, the total dose, the total number of MUs, and the equivalent square field sizes as well as average neutron-equivalent (H_PN) and out-of-field doses (D_OOF) per MU and treatment fraction are presented in Table 2 as mean values and standard deviations. Comparing all respective plans together, it becomes apparent that over all of the different scenarios, SIB resulted only in 57 plans in more MUs but actually in another 63 plans in less MUs than their CONV counterparts. Furthermore, the resulting neutron-equivalent—and out-of-field doses per MU and per complete treatment as well as the correlation coefficients are displayed.

This study is aimed to investigate whether implementation of a simultaneous integrated boost (SIB) results in an effect on neutron or scattered photon doses, possibly due to increased MLC activity, which has been identified as one primary source for neutron generation [10]. Furthermore, we explored whether the FFF mode may influence such an effect, if present.

In our 240 treatment sequences, the following case-dependent results were found: In 115 (for D_OOF) and 118 (for H_PN) out of 120 treatment courses, implementation of an SIB resulted in less D_OOF and H_PN when compared to CONV. The FFF mode was preferable over FLAT in all cases with respect to D_OOF. For 6 MV, neutron doses were higher for FFF beams in all cases (60/60). For 10 MV, neutron doses were lower in 54 of 60 cases with the FFF mode. As expected, neutron doses were always higher for 10 MV than for 6 MV. On the other hand, D_OOF values were lower in 104 of the 120 cases for 6 MV.

The SIB concept leads to an increased fractional RT dose within a selected volume. When calculated for comparable efficacy, the EQD_2Gy model calculates for a given alpha/beta value a reduced total RT dose, which in turn implies a reduction in the fraction number, therefore shortening the overall treatment time. We found that our SIB scenario (2.5 Gy/3.0 Gy to 50 Gy/60 Gy) results in decreased total neutron dose as well as D_OOF when compared to conventional (CONV) fractionated 2 Gy to 74 Gy. Therefore, due to the higher fractional doses and thus reduced number of treatment fractions, an SIB is clearly preferable in this context. Flattening filter-free VMAT, effective in sparing both neutron dose (in 10-MV beams) and D_OOF, respectively, increased this effect.

Omission of the flattening filter (FFF) significantly decreases both the photoneutron as well as the out-of-field dose in VMAT for prostate cancer at distant locations, even though FFF plans typically require more MUs than their flattened counterparts to compensate for the non-flatness of its profiles. Monitor units as surrogate for the modulation degree have a strong linear correlation with the formation of photoneutrons. By contrast, with respect to out-of-field dose, this strong linear correlation with MUs was only noted in flattened SIB plans.

We show that no increased neutron (H_PN) or scattered photon dose (D_OOF) results from increasing the fractional dose from 2.0 Gy to 2.5/3.0 Gy. Therefore, implementation of an SIB in a hypofractionated dose concept leads to less neutron and out-of-field dose in a complete treatment course and this effect was as distinct as the neutron dose-sparing and scattered radiation-sparing effect exerted by FFF VMAT.

Since the measurement position was kept constant at all times at 61.8 cm away from the beam isocenter, we were unable to detect a relationship between the distance to the radiation field and scattered dose. Hauri et al. [22] demonstrated a dependency of scattered dose on the distance from the isocenter. Between 40 and 80 cm away from the isocenter, total doses decrease slowly and the resulting scatter dose results mainly from collimator scatter and head leakage (and less from patient scatter). Haelg et al. [23] showed that neutron contamination depends only weakly on distance from the radiation field. Thus, a comparison between different treatment plan parameters (6 vs. 10 MV, FFF vs. FLAT, and SIB vs. CONV) appears to be appropriate at a fixed distance and the results presented may be used to approximate a whole-body exposure.

In our scenarios the boost dose is 3.0 Gy while the surrounding PTV receives 2.5 Gy, hence a ratio of 1.25 to 1.5 when compared to CONV treatment plans (2 Gy). For the sake of comparison, all absolute neutron and out-of-field dose readings were taken in relation with MUs to account for these differing dose levels. When analyzing all 30 patients, we found both increased and decreased numbers of MUs for the respective SIB plans compared to their corresponding CONV treatment plans. When the 120 SIB treatment plans were corrected by 3 Gy/2.5 Gy (to yield MU per Gray PTV dose), 57 SIB treatment plans had more MUs and 63 SIB plans exhibited less MUs than their CONV counterparts. Non-uniform target coverage therefore requires in every second case less MUs than their uniform equivalents. This also implies less neutron dose.

Due to the higher number of MUs required in FFF dose deliveries, the reduced neutron generation has to be set in relation to the difference in MUs to judge whether FFF or FLAT deliveries generate less neutron dose to the patient. In flattened SIB plans (6 MV and 10 MV), we noticed a comparable correlation between MUs and D_OOF, an effect which was circumvented by the omission of the FF.

In AAPM report 158 [12], deposited neutron dose equivalents per MU for 6‑MV and 10-MV Elekta linacs are not provided. However, our finding of 0.5 µSv/MU for 10-MV FLAT are in line with the reported Varian 10-MV FLAT beam results of 0.9 µSv/MU and 0.3 µSv/MU for Siemens linacs. It is, moreover, noted in AAPM report 158 that Varian linacs possess a roughly twofold higher neutron contamination than comparable Elekta linacs.

Even though no relevant neutron dose was measured in 6‑MV beams, a stable signal was obtained at a very low level, which also showed a correlation between MUs and H_PN with r values between 0.93 and 1.0 (Table 2), and a slight increase in average neutron dose was noted for 6‑MV FFF when compared with 6‑MV FLAT. Of note, even for this beam energy, a reduction in H_PN is induced for an overall RT course with implementation of SIB (34.7 and 37.1% for 6‑MV FLAT and 6‑MV FFF, respectively; Table 3).

At the nominal energy of 6‑MV FLAT, the flattening filter does not contribute to PN contamination, because the bremsstrahlung is filtered with a stainless steel disc [24] and energy thresholds for PN production (γ, n) for iron and copper are 11.2 MeV (91.7%, Fe-56), 10.9 MeV (69.2%, Cu-63), and 9.9 MeV (30.3%, Cu-65; [25]). The same IAEA report lists all energy thresholds for PN production (γ, n) for the respective isotope and its natural abundance for tungsten. These were reported to be 8.1 MeV (26.3%, W‑182), 7.4 MeV (30.7% W‑184), 6.2 MeV (14.3%, W‑183), and 5.8 MeV (28.6%, W‑186). The 6‑MV Elekta linac’s maximum photon energies were reported to be 6.6 or 6.7 MeV, respectively [26, 27]. Furthermore, Elekta linacs—in contrast to Varian linacs—increase the initial electron energy for FFF beams up to a point where the D10cm of FLAT and FFF is identical (e.g., 67.5/73.5% of the maximum dose at 10 cm depth for 6/10 MV; [28]). This, in summary explains why detected Hpn values for 6‑MV FFF are higher than 6‑MV FLAT: 0.027 µSv/MU > 0.015 µSv/MU. For 10 MV this should apply as well; however, the flattening filter as an additional neutron source dominates and reverses the behavior 0.5 µSv/MU (10-MV FLAT) > 0.3 µSv/MU (10-MV FFF).

In our study, the SIB scenario reduced for 10-MV plans the total neutron dose from 13.1 to 8.4 mSv (36.1%) in flattened VMAT and from 10.9 to 6.5 mSv for FFF-VMAT (40.2%). FFF-VMAT reduced neutron dose from 13.1 mSV to 10.9 mSV (17.2%) for the conventional fractionated scenario and from 8.4 to 6.5 mSv (22.5%) for SIB-RT. Scattered radiation (D_OOF) was likewise reduced for both 6‑MV and 10-MV beams with SIB by approximately 36.1–40.9% and even more effectively with FFF by approximately 59.1–61.5%. The lowest neutron and scattered radiation dose was derived with 6‑MV SIB (H_PN 0.3–0.6 mSv; D_OOF 14.0 mGy for 6‑MV FFF).

Treutwein et al. [29] compared the excess absolute risk (EAR) of secondary cancer in localized prostate IMRT with 6‑MV beams with and without flattening filter. For the secondary malignancy risk (SMR) in the periphery—corresponding to our distant measurement localization—they noted a reduced peripheral dose without flattening filter, which led to significantly reduced SMR. Murray et al. [13] described lower second cancer risks for FFF with increasing impact in organs at greater distance. This beneficial effect may be increased with SIB-RT. These reported SMR induced by scattered photons and electrons are very low for regions in the body at a distance from the primary treated region. Nevertheless, photoneutrons pose a significant risk in view of secondary cancers (ICRP 60; [30]) and even more so to implanted electronic devices because of their interaction with complementary metal oxide semiconductors and random access memory [31‐33]. Hauri et al. presented data that showed that neutron dose equivalent remained unaffected by the distance from the isocenter and neutron dose peaked in the neck region along an axis from the pelvis (isocenter) to the head in an Alderson phantom, probably due to different thickness of penetrated tissue [22]. Corresponding to our data, Haelg et al. showed that neutron dose resulting from 15-MV IMRT produced by linacs from three different vendors (Elekta, Varian, and Siemens) was at least one order lower than the photon stray dose, therefore contributing with a very low amount to the total integral dose of a patient [23]. Still, regarding neutron dose, significant effects were noted in patients and phantom studies with respect to cardiac implanted electronic devices (CIEDs, [32, 33]). These devices are implanted near the neck region under the clavicular bone in a subcutaneous pouch, therefore in the peak region for neutron dose [22]. In a phantom study, 10-MV FFF-VMAT of the prostate resulted in pacing inhibition, data loss and reset in 4 of 15 ICDs that were located at a distance from the primary beam [19]. In an exemplary clinical survey, 10-MV FFF-VMAT resulted in an ICD error [32]. Other clinical studies demonstrate that photon beams >6 MV result in CIED errors [31‐33] even though beam energies were widespread between 10 and 23 MV. International guidelines advocate the use of non-neutron-producing beam energies in CIED-bearing patients, but discrepancy exists between a lower beam energy threshold (6 vs. 10 MV, [18, 31, 33‐35]). No data exist for the disaggregation between 10-MV FLAT and 10-MV FFF in regard to CIED effects even though we have shown a 20% reduction in neutron dose in 10-MV FFF VMAT when compared to their 10-MV FLAT counterpart plans (for the same fx dose level). Therefore, although we show that SIB can further reduce H_PN in FFF-VMAT (H_PN 10-MV FLAT > 10-MV FFF > 10-MV FLAT SIB > 10-MV FFF SIB), we cannot advise for the use of 10-MV FFF beams even for RT with SIB in CIED-bearing patients and have to wait for further research on this topic.

Our study has some limitations that need consideration. We only used one measurement point approximately 60 cm from the field edge 0.5 cm below the surface of the phantom. We chose this location with reference to Hauri et al. [22]. At 60 cm distance, there is almost exclusively head leakage contribution and almost no patient scatter. Thus, past this distance the measured doses remain approximately constant. Closer to the radiation field, both Hauri et al. and AAPM report 158 [12] report an approximately exponential decay of patient scatter and a constant head leakage contribution. Furthermore, this distance between detectors and radiation field reflects clinically the distance between prostate RT and CIEDs [19] and considers H_PN and D_OOF at their typical location in the body.

In conclusion, implementation of a simultaneous integrated boost (SIB) and omission of the flattening filter (FFF) reduces neutron contamination and out-of-field doses per monitor unit for volumetric modulated arc therapy (VMAT) for prostate cancer. SIB treatments lead to lower neutron and out-of-field doses for an overall treatment course. Neutron contamination in VMAT for prostate cancer depends, as expected, predominantly on the beam energy and the number of delivered monitor units. Due to the higher number of monitor units typically required in flattening filter-free (FFF) dose deliveries with this particular set-up, the reduced neutron generation per monitor unit has to be set in relation to the difference in monitor units to judge whether FFF or FLAT deliveries generate a lower neutron and out-of-field dose to the patient. Respecting radiation protection principles, an implementation of an SIB in moderately hypofractionated prostate RT may be advisable not only for shortening of overall treatment time.