Calculating and estimating second cancer risk from breast radiotherapy using Monte Carlo code with internal body scatter for each out‐of‐field organ

Abstract Out‐of‐field organs are not commonly designated as dose calculation targets during radiation therapy treatment planning, but they might entail risks of second cancer. Risk components include specific internal body scatter, which is a dominant source of out‐of‐field doses, and head leakage, which can be reduced by external shielding. Our simulation study quantifies out‐of‐field organ doses and estimates second cancer risks attributable to internal body scatter in whole‐breast radiotherapy (WBRT) with or without additional regional nodal radiotherapy (RNRT), respectively, for right and left breast cancer using Monte Carlo code PHITS. Simulations were conducted using a complete whole‐body female model. Second cancer risk was estimated using the calculated doses with a concept of excess absolute risk. Simulation results revealed marked differences between WBRT alone and WBRT plus RNRT in out‐of‐field organ doses. The ratios of mean doses between them were as large as 3.5–8.0 for the head and neck region and about 1.5–6.6 for the lower abdominal region. Potentially, most out‐of‐field organs had excess absolute risks of less than 1 per 10,000 persons‐year. Our study surveyed the respective contributions of internal body scatter to out‐of‐field organ doses and second cancer risks in breast radiotherapy on this intact female model.

conventional radiotherapy techniques. 5 Contributions of head leakage and collimator transmission radiation are described in the literature. [5][6][7] Earlier findings suggest that the magnitude of leakage head scatter that contributes to the out-of-field dose can be reduced by placing a lead shield over the critical area. 8,9 Internal body scatters, however, cannot be shielded. Knowledge of exposure attributable to internal body scatter is crucially important for radiation risk management. Nevertheless, measuring the dose directly in each organ in a patient body is impossible.
Out-of-field organ doses must be managed to mitigate second cancer risk, just as they are for in-field organs, but they are not calculated in general radiotherapy treatment planning. Moreover, treatment planning system (TPS) accuracy is usually limited to a few centimeters outside the field edge. 10,11 Earlier studies have estimated out-of-field absorbed doses using anthropomorphic phantom measurements or TPS calculations on patient computed tomography (CT) images. Kry et al. calculated doses for out-of-field organs and estimated second cancer risks after prostate radiotherapy using TPS. 12 Howell et al. evaluated the accuracy of TPS for out-of-field dose calculation compared to measurements using an anthropomorphic phantom. 10 Zeverino et al. summarized the peripheral organ dose in breast radiotherapy depending on the planning technique for several TPSs. 13 Kourinou et al. measured the contribution of internal body scatter to the fetal dose using an anthropomorphic phantom. 14 These experimental reports have described approximate doses in out-of-field organs in TPS for breast cancer, but out-of-field organ doses have not been established based on physical simulation with a complete whole-body model. Some refined analytical models and systems are able to predict out-of-field doses. 15,16 However, scattering in the complex structure of the human body is not considered. Also, the dose of a specific organ is not a calculation target.
Monte Carlo (MC) simulation, which is well known as a physically accurate dose calculation method, has been used in some earlier studies to calculate out-of-field absorbed doses. Reda et al. used a simple anthropomorphic phantom to calculate the doses for some organs for one situation of breast radiotherapy. 17 Joosten et al. calculated out-of-field organ absorbed doses by simulating irradiation of left breast radiotherapy on a CT-based female phantom that did not include forearms or lower limbs. 18,19 Abo-Madyan et al. estimated the second cancer risk for the breast and lung after breast radiotherapy. 20 However, no report of the relevant literature describes a study validating irradiation effects for regional lymph nodes from a low-dose bath or a study estimating the second cancer risk for out-of-field organs mainly attributed to internal body scatter in radiotherapy for the breast in a complete whole-body MC simulation.
This study specifically undertook simulation of all organ absorbed doses by internal body scatter from conventional radiotherapy techniques using WBRT alone and WBRT plus RNRT, with calculation of second cancer risks in these radiotherapies using the concept of the organ equivalent dose (OED) 21 24 for postural modification of the model is combined with the whole-body model package. The model posture was modified for dose calculation: it was set to resemble a realistic position of raising both arms above the head (Fig. 1). We assigned mass density and compositions for all tissues and organs based on the International Commission on Radiological Protection (ICRP) publication 110 25 and allocated air around the model.

2.B | Treatment planning
The WBRT plans consisted of two opposing tangential fields with two additional opposing subfields for the breast using multileaf collimators (MLCs), that is, "field-in-field technique." RNRT plans had two opposing regional lymph node fields using MLCs. The prescribed dose of WBRT is 50 Gy to the breast in 25 fractions. That of RNRT is 50 Gy to regional (subclavicular and internal mammary) lymph nodes in 25 fractions. For treatment planning, a combination of 4-and 6-MV x-ray beams was used. Above is a standard of care for breast cancer patients at our institution. The convolution-super-   3.02. 23 PHITS incorporates Electron Gamma Shower Version 5 (EGS5), in which photon and electron transport were established as valuable tools to simulate a medical linac. 26,27 As incident x-ray spectra, we used MC modeled photon fluence spectra reported by Shiekh-Bagheri and Rogers. 28 Beam information such as the leaf position, jaw position, and beam angle were obtained from DICOM RT files planned in Pinnacle 3 TPS for the MC simulation. The whole-body voxel model was downsampled to voxel size 2 × 2 × 4 mm 3 (coronal, axial, sagittal) to input the model into PHITS and to reduce statistical uncertainty. To simulate only the dose contributed from internal body scatter, which is the principal component of out-of-field doses, 4,5 scattered radiation produced by the linear accelerator, for example, head scatter, collimator transmission, and MLC transmission, was not modeled for dose estimation in this study. To validate the beam source models, we compared the percentage depth dose (PDD) and beam profiles of 4-and 6-MV x-ray beams in water (50 × 50 × 50 cm 3 ) for 10 × 10 cm 2 field and 100 cm source-to-surface distance between the PHITS simulation and TPS.
The photon and electron cutoff energies were, respectively, 10 keV and 100 keV. The photon mean free path at 10 keV in liquid water is approximately 1.9 mm. 29 The electron attenuation length at 100 keV in liquid water is approximately 0.14 mm. 30 They were smaller than the voxel size.
Two billion incident photons for each simulated beam were used to achieve statistical uncertainties of less than 2% at the target breast conclusively. Generally, MC simulation requires a long duration and tremendous computational power to reduce statistical uncertainty. We accelerated simulation by parallel computing using a supercomputing system (Reedbush-H, two 18 core processors, Xeon

2.D | Calculation of second cancer risk
We estimated the risk of second cancer by the excess absolute risk (EAR) optimized against the epidemiological data. The cancer incidence is assumed to be directly proportional to OED, according to the linear no-threshold model. Therefore, EAR after radiotherapy is calculated as.
where EAR T stands for EAR after radiotherapy for organ T (per 10,000 person-year (PY)), EAR T0 denotes epidemiological EAR, and OED T signifies OED for organ T. 20 EAR derived from Japanese atomic bomb survivor data for organ T (per 10,000 PY at 1 Gy at the age of 70 yr, following exposure at the age of 30 and 50 yr), as described by Preston et al. 32 The OEDs for most out-of-field organs were calculated using the linear dose-response model mainly because the model accomplished Whole-body female model designed by the National Institute of Information and Communications Technology. 22 Fat, muscle, blood, and skull are not visible in this figure..
good precision under 2 Gy. 33 Most out-of-field organs in radiotherapy were exposed to less than 2 Gy in this study. Therefore, we calculated OED for the linear dose-response model as shown below.
That equation shows that, for each dose level, the summation from i equals zero to n of dose D in organ T, which has volume V T .
Also, V Di is the volume corresponding to dose D i . 33 The mean absorbed doses for the target breast, the ipsilateral lung, the thyroid, the skin, the large intestine, the liver, the esophagus, and the spleen were sometimes found to be over 2 Gy in our study. For doses greater than 2 Gy, the dose-response relation for cancer is no longer linear. This model would generally overestimate the OED. 34 It has often been hypothesized that the linear dose-response model is no longer applicable because of cell death at higher doses. The linear-exponential dose-response model is proposed as one model to describe the dose-response at higher doses, as explained below. 21,33 Therein, model parameter α was estimated from a combined fit to the Japanese atomic bomb and Hodgkin cohorts. 21,34 3 | RESULTS  absorbed dose for the target breast was less than 1% between PHITS and TPS (convolution-superposition algorithm) in all treatment plans.
The dose volume histograms (DVHs) for the target breast and organs near the radiation field by PHITS and TPS are shown in Fig. 4. The gamma index passing rate for them is shown in Table 1.
The simulated dose for the target breast accorded well with TPS.
Earlier reports of experimental studies have described that the convolution-superposition algorithm in TPS apparently overestimates the absorbed dose for the out-of-field organs around the chest. 35-37 Figure 5 and Fig. 6 present part of DVHs for out-of-field organs simulated using PHITS (Simulated absorbed doses presented in Table 2).
Full MC simulation was useful to survey the absorbed dose distribution from head to toe, although distant organs such as the genital  Comparison of out-of-field organ doses between WBRT alone and WBRT plus RNRT plans revealed that the latter was inclined to increase the maximum dose for out-of-field organs. Also, RNRT produced the mean dose increase in contrast to WBRT alone. The ratios of mean doses between WBRT alone and WBRT plus RNRT were approximately as large as 3.5-8.0 at the head and neck region and around 1.5-6.6 in the lower abdominal region. This increase resulted from scattering radiation which proceeded via regional lymph node fields. Table 3 summarizes EAR for out-of-field organs (per 10,000 PY at the age of 70 yr after exposure at the ages of 30 and 50 yr) for conventional radiotherapy techniques calculated using Eqs.
(1) and (2). Additional RNRT increased to the same degree as that for EAR found for WBRT only. EAR of the stomach with WBRT plus RNRT for the left breast was 13.2, which was the highest value excluding the breast and lung. Other organs with EAR higher than 1 are the thyroid and esophagus which were reported to increase second cancer risk. Furthermore, the EARs of the large intestine and liver were also higher than 1. The EAR of other out-of-field organs was less than 1.

| DISCUSSION
This study simulated absorbed doses for 43 organs and estimated the risk of second cancer for major organs attributed to specific internal body scatter from breast radiotherapy with WBRT alone and WBRT plus RNRT using full MC code PHITS in the complete wholebody female model. The simulated mean absorbed dose for the target breast matched with TPS within 1%. Substantial differences were found, however, between them for the low-dose bath such as the lung and heart (Fig. 4). In that regard, the estimated uterus and ovary doses were consistent with reported doses because of internal body scatter at the fetal level from breast radiotherapy with two opposing tangential fields. 14 Comparison of the out-of-field organ doses between those found for WBRT alone and WBRT plus RNRT indicates that the latter greatly increased the absorbed doses for most out-of-field organs.
As one might expect, the exposure dose for organs around the head and neck increases because of opposing regional lymph node fields in RNRT. Furthermore, these fields markedly increased the dose for all out-of-field organs including the lower abdomen. Although these additional fields did not overlap with opposing tangential fields for the breast, the scattered radiation to out-of-field organs originated from each field was not negligible.
Modern photon radiotherapy techniques such as intensity-modulated radiation therapy (IMRT) and volumetric modulated arc therapy described dose overestimations around the lung caused by the scatter kernel of the convolution-superposition algorithm. [35][36][37] The difference between PHITS and TPS might have been caused by the treatment of secondary electron components in low-density inhomogeneous media. The lateral electronic equilibrium is difficult to model because the range of secondary electrons in low-density inhomogeneous media, such as the air and lung, is longer than that in water.
By contrast, MC simulation in this study was conducted with actual elemental compositions that are allocated directly to each tissue on the whole-body model. Therefore, the MC simulation algorithm is expected to be valid for estimation of a low-dose bath including regions that are distant from the target.
The contribution of internal body scatter over lead leakage has been reported in the literature. 14 Nevertheless, no report of the literature presents data for specific organ doses and second cancer risks posed by internal body scatter in breast radiotherapy. Our simulation results demonstrate clearly that unavoidable EAR for out-offield organs was less than 1. This result indicates that the risk of second cancer is much lower than for the breast, lung, thyroid, and esophagus which were earlier reported to show increased risk. The risk might be negligible for these organs. However, the skin on the surface of the human body might be more affected by head leakage, collimator transmission, and room scattering from all directions than the internal body organs. Evaluating the second cancer risk for the skin requires further investigation. Although the risk is small, WBRT plus RNRT elevates the risk of the second cancer of out-of-field organs considerably, by as much as 3.5-8 times for the head and neck region, compared to WBRT according to the dose calculation conducted for this study. That risk is nonetheless acceptable because radiotherapy benefits outweigh the attendant risk.
Medical safety demands our best efforts to reduce this risk. The relation between radiation therapy and second cancer is being clarified. 3,44,45 Combining the specific organ doses obtained in this study with prognosis of the patient is expected to clarify the causal relation between the absorbed dose and second cancer in breast radiotherapy.
Regarding the effect of uncertainty in dose estimation on estimating the risk of second cancer, reports of the relevant literature claim that 50% uncertainty in the calculated peripheral dose is acceptable given the uncertainties of risk factors for second cancers. 46 However, the risk of second cancer might be different when a dose uncertainty of at least 50% exists. 47 In our study, the statistical uncertainties of out-of-field organ doses were mostly less than 50%. Therefore, assuming 50% as the threshold of dose uncertainty for risk estimation based on the former claim, then the statistical uncertainties of organ doses which have less than 50% uncertainty can be acceptable for risk estimation in our study. However, the uncertainties for some organs such as the bladder, uterus, and ovaries exceeded 50%. The estimated risks of second cancer for these organs might be inaccurate based on the assumption, but the risks would still be sufficiently small as to be negligible. The linear-exponential dose-response model was used for organs exposed over 2 Gy. The linear dose-response model was used for other organs. EAR based on Japanese atomic bomb survivor data is not reported if no value is shown. Values in parentheses are 90% confidence intervals.
type has little effect on a low-dose bath. 48 However, DVHs for peripheral organs differ considerably among patients. Treatment plans such as different irradiation of right and left breasts were examined in this study. Therefore, when estimating the risk of second cancer for peripheral organs, they are calculable with the patient's individual CT dataset used during treatment planning.
A second limitation is the coverage of Japanese atomic bomb survivor data. Marked differences exist in the rates of some cancers, such as breast and ovary cancer, between the general Japanese population and the world population. 49 This striking interpopulation difference in baseline rates for some cancers presents a complicated issue, as discussed in earlier reports. 50,51 Moreover, the EAR data include no distinction between men and women. Therefore, it is difficult to apply the data to the rates of some cancers for which gender differences are large, such as bladder cancer. Extrapolating the results to the world population is hindered by these inherent limitations.

| CONCLUSION
We applied MC dose simulation and estimated the second cancer risks for a low-dose bath by internal body scatter for WBRT alone and WBRT plus RNRT with realistic radiotherapy conditions using a T.K., and J.K were involved in analysis, or interpretation of data..

CONF LICT OF I NTEREST
The authors have no relevant affiliations or financial interests to report.