Second cancer risk after radiation therapy of ependymoma using the flattening filter free irradiation mode of a linear accelerator

Abstract Pediatric patients suffering from ependymoma are usually treated with cranial or craniospinal three‐dimensional (3D) conformal radiotherapy (3DCRT). Intensity‐modulated techniques spare dose to the surrounding tissue, but the risk for second malignancies may be increased due to the increase in low‐dose volume. The aim of this study is to investigate if the flattening filter free (FFF) mode allows reducing the risk for second malignancies compared to the mode with flattening filter (FF) for intensity‐modulated techniques and to 3DCRT. A reduction of the risk would be advantageous for treating pediatric ependymoma. 3DCRT was compared to intensity‐modulated radiation therapy (IMRT) and volumetric‐modulated arc therapy (VMAT) with and without flattening filter. Dose–volume histograms (DVHs) were compared to evaluate the plan quality and used to calculate the excess absolute risk (EAR) to develop second cancer in the brain. Dose verification was performed with a two‐dimensional (2D) ionization chamber array and the out‐of‐field dose was measured with an ionization chamber to determine the EAR in peripheral organs. Delivery times were measured. Both VMAT and IMRT achieved similar plan quality in terms of dose sparing in the OAR and higher PTV coverage as compared to 3DCRT. Peripheral dose in low‐dose region, which is proportional to the EAR in organs located in this region, for example, gonads, bladder, or bowel, could be significantly reduced using FFF. The lowest peripheral EAR and lowest delivery times were hereby achieved with VMATFFF. The EAR calculated based on DVH in the brain could not be reduced using FFF mode. VMATFFF improved the target coverage and homogeneity and kept the dose in the OAR similar compared to 3DCRT. In addition, delivery times were significantly reduced using VMATFFF. Therefore, for radiotherapy of ependymoma patients, VMATFFF may be considered advantageous for the combination of Elekta Synergy linac and Oncentra External Beam planning system used in this study.


| INTRODUCTION
The standard therapy for childhood ependymoma is surgery followed by adjuvant radiation. 1,2 Historically, cranial or craniospinal irradiation was the standard treatment after surgery 3 in order to assure that the volume adjacent to the target receives adequate dose and consequently assure local control. However, recent publications have shown that conformal, intensity-modulated (IMRT) and proton-beam radiotherapy can achieve similar local control rates to those published in historical publications without an increased risk of marginal failure. [4][5][6][7] An advantage of IMRT over three-dimensional (3D) conformal standard technique is to spare dose in surrounding normal tissues.
However, a concern about the IMRT treatment is that the volume that receives a low dose can be significantly higher compared to those volumes for conventional technique. This might increase the risk for second malignancies, which is a very important issue with regard to the life expectancy of these pediatric patients. 8 A recent development in the linear accelerator technology is the irradiation without a flattening filter in the beam path to increase dose rate and reduce beam-on time. 9 A reduction of the treatment time would represent an additional advantage, especially in the treatment of pediatric patients. The flattening filter produces scatter radiation; therefore, its removal has the positive effect of reducing the out-of-field dose, 10 which may lead to reduced second cancer risk as it has been shown for the treatment of breast cancer in a previous study. 11 To our knowledge, no reports about second cancer risk have been published for ependymoma. The excess absolute risk (EAR) of developing a second cancer after exposure to radiation (EAR) can be estimated from dose-volume histograms (DVHs) based on biological models which are fitted to data of atomic bomb survivors and Hodgkin patients treated with radiation therapy. 12,13 The purpose of our study was therefore to investigate whether the flattening filter free mode (FFF) is advantageous with respect to the second cancer risk, plan quality, and delivery time after radiation therapy of ependymoma as compared with the flat beam mode (FF).

2.A | Patients and treatment planning
Computed tomography (CT) data of 11 pediatric patients with ependymoma were selected from our treatment database for a retrospective planning study. The age of the patients ranged from 22 months to 15 yr with a mean of 6.5 yr and a median of 4.6 yr. All patients were completely resected and without metastases. The irradiation approach was used in this planning study in the same way as in the HIT Interim Register for children under 4 yrs 14 (dose per fraction 1.8 Gy, total dose 50.4 Gy). HIT stands for brain tumors in children. The precursor was the HIT 2000 study, a multicenter therapy optimization study, which provided optimal and risk-adapted therapy nationwide in Germany and Austria for patients up to the age of 21 with an intracranial localized medulloblastoma, PNET, or ependymoma. Following the completion of the HIT 2000 study, the HIT Interims Register study was launched to bridge the period until the relaunch of the study with the same standard of quality.
The patients were in supine position and an individual thermoplastic facial mask was created. The planning CT was then performed in 4 mm layer thickness from occiput to the third cervical vertebral body in a multislice CT Siemens ® Somatom Sensation Open. The planning CT was imported into the planning system and it was fused with the magnetic resonance tomography (T1-weighted, contrastenhanced MRI) before and after surgery (MRI equipment: 1.5T Siemens ® Magnetom Avanto and Siemens ® Magnetom Symphony). The planning target volume (PTV) was the expanded tumor region, which means the former tumor region (GTV = gross tumor region), including two centimeter of automatically generated safety margin taking into account anatomical limits or rearrangement of brain tissue into the resection area after surgery. Dose threshold was based on the tolerance doses for organs at risk (OAR) of the protocol of Radiation Therapy Oncology Group Trial 0225. 15 Dose to the normal tissue (patient outline excluding the PTV) should be kept as low as possible to minimize the risk of second cancer induction. 8 Treatment planning was performed with Oncentra External Beam  Table 1. DVOs to the OAR were set to values which could be achieved in 3DCRT. Identical DVO were used for optimization of all plans. The dose calculation was performed with a grid size of 3 mm and the dose was normalized to the average dose in the PTV.
All plans were accepted for treatment by a radiation oncologist.  (bottom) and a 9.7 cm (top) stack of RVW slabs (PTW, Freiburg, Germany). AAPM TG119 recommendations were used for the dose verification acceptance. 16,17 Gamma indices 18 were calculated with a dose tolerance of 3% of the maximum dose and 3 mm distance to agreement and the gamma criterion was considered fulfilled if γ < 1 in at least 95% of the pixels.
In the low-dose region, the EAR presents a linear dose-response. 19,20 Since planning CT data do not include images at larger distances for reasons of radiation protection, no calculations of the EAR based on DVH are possible, for example, bladder, bowel, and gonads. Various studies have previously shown large uncertainties for dose calculations in the out-of-field area and therefore recommended to perform point dose measurements instead of calculations in this area. 9,21,22 Therefore, we performed dose measurements in the same coronal plane in the low-dose region at a distance of 31 cm cranial of the isocenter using a 0.3 ccm ionization chamber (PTW) simultaneously with 2D verification (Fig. 1). The target is symmetric in cranial-caudal direction; therefore, the measurements can also be associated with the dose in the caudal region. The results were compared for the five irradiation techniques for the whole series. The complete setup is shown in Fig. 1.
Simultaneously, total delivery times were measured from first beam on to last beam off.

2.C | Excess absolute risk calculation and organ equivalent dose
The excess absolute risk (EAR) to develop a solid cancer describes the absolute difference in cancer rates of persons exposed to a dose d and those not exposed to a dose beyond the natural dose exposition per 10,000 person-year and is described as 23 EARðD; e; aÞ ¼ EAR 0 REDðDÞ lðe; aÞ where RED (risk equivalent dose) is the dose-response relationship for radiation-induced cancer in units of dose and EAR 0 the initial slope (slope of the dose-response curve at low dose). The function μ depends on the attained age a and age at exposure e: For this study, the EAR values are calculated for gender-averaged persons at an age at exposure of 10 yr and an attained age of 70 yr and reported per 10,000 person-years. EAR 0 , γ e , and γ a values have T A B L E 1 DVO used for optimization and tolerance dose for each structure.
been derived by Preston et al. 20 from the data of the Japanese atomic bomb survivors (EAR 0 = 0.7 per 10,000 person-years per Gy for the brain, γ e = −0.024, and γ a = 2.38). Large errors are involved in the determination of the parameters EAR 0 and μ; therefore, absolute risk results have to be viewed with care.
It is known, that for doses below 2 Gy, the dose-response is linear. 8 For higher doses, the concept of an organ equivalent dose (OED), which is proportional to EAR, has been defined by Schneider et al. 12,13 : Therefore, risk ratios for different treatment plans are equivalent to OED ratios which can be determined with the RED function and DVH.
There are different models for the RED calculation, based on different assumptions on the behavior of cells after dose exposition. 23 The linear model assumes a linear response over the whole dose range: REDðDÞ ¼ D The mechanistic model accounts for cell killing and fractionation effects 23,24 : where / 0 ¼/ þb D DT d T For this model, there are two limit cases: when the parameter R is 0 if no and 1 if full repopulation/repair occurs. Therefore, the limit R = 0 leads to the linear-exponential model: REDðDÞ ¼ DexpðÀa 0 DÞ Moreover, the limit R = 1 is the case of full repopulation/repair, known as plateau model: The parameters α and β are those from the linear quadratic model of cell killing. The parameter R describes the repopulation and repair ability between dose fractions. The parameters α′ = 0.018 and R = 0.93 for the brain have been derived from a combined fit to the data of atomic bomb survivors and Hodgkin patients treated with doses up to 40 Gy assuming that α/β = 3 Gy. 23 α′ is 0.009 in case of no fractionation (linear-exponential limit) and 0.021 in case of full tissue recovery (plateau limit).
As mentioned before, the parameters used in the calculation of absolute EAR involve large errors. In order to keep the uncertainties at minimum, we opted to use the OED values for evaluation of the risk of second cancer when comparing different radiation techniques. EAR absolute values were also calculated for the sake of completeness.
Previous studies have demonstrated that for inhomogeneous dose distributions above 4 Gy, the linear-exponential, the plateau, and the full mechanistic model represent a better description of the dose-response function compared to the linear model. 12,23 Therefore, this model was not included in our results. Plan quality was evaluated by means of the DVH and dose distributions. Details of relevant DVH parameters are given in Table 2 averaged over all patients. Threshold doses listed in Table 1

3.A | Dosimetry and delivery
Gamma evaluation of the verification of dose calculation is shown in Table 3. All 44 IMRT and VMAT plans passed the evaluation with passing rates >95% as recommended by the AAPM TG119. 16,17 On average, IMRT FFF achieved a higher gamma passing rate than the other techniques, but no significant differences were found.
Total delivery times are also listed in Table 3. Both VMAT modes reduced the delivery time significantly (P = 0.003) by 50% compared to 3DCRT. Both IMRT modes doubled significantly the delivery time compared to 3DCRT (P = 0.003).
Peripheral dose was measured with an ionization chamber in the measurement setup of Fig. 1; the measured doses are listed in

| DISCUSSION
The main objective of this study is to investigate the potential of the FFF mode of a linear accelerator to reduce the radiation-induced second cancer risk, the treatment time, and to improve the treatment plan quality in pediatric ependymoma.
The major concern of using inverse treatment planning for the treatment of ependymoma is that the high-dose falloff at the margin of the PTV might lead to a decrease of the local tumor control.
However, recent publications for ependymoma showed that the local control and overall survival achieved with IMRT are not inferior to those associated with conventional therapy. 4 therefore, all significant results remained significant after the test.

| CONCLUSION
The results of this study show that VMAT FFF is a technique to consider for treating childhood ependymoma. Compared to the standard technique used for ependymoma, VMAT FFF achieved significantly higher PTV coverage and allowed reducing treatment time significantly. The risk of radiation-induced secondary cancer, which is a major concern in radiation therapy of pediatric patients, was significantly reduced for organs located at large distance from the target, for example, bladder, gonads, or bowel. No advantage could be observed with respect to the secondary cancer risk in the brain.

ACKNOWLEDG MENTS
This study was funded by the Bavarian State Ministry of the Environment and Consumer Protection. The publication was supported by the German Research Foundation (DFG) within the funding program Open Access Publishing.

CONFLI CT OF INTEREST
The authors declare that there is no conflict of interest regarding the publication of this article.