Estimation of secondary cancer risk after radiotherapy in high‐risk prostate cancer patients with pelvic irradiation

Abstract We aimed to estimate the risk of secondary cancer after radiotherapy (RT) in high‐risk prostate cancer (HRPC) patients with pelvic irradiation. Computed tomography data of five biopsy‐proven HRPC patients were selected for this study. Two different planning target volumes (PTV1 and PTV2) were contoured for each patient. The PTV1 included the prostate, seminal vesicles, and pelvic lymphatics, while the PTV2 included only the prostate and seminal vesicles. The prescribed dose was 54 Gy for the PTV1 with a sequential boost (24 Gy for the PTV2). Intensity‐modulated RT (IMRT) and volumetric modulated arc therapy (VMAT) techniques were used to generate treatment plans with 6 and 10 MV photon energies with the flattening filter (FF) or flattening filter‐free (FFF) irradiation mode. The excess absolute risks (EARs) were calculated and compared for the bladder, rectum, pelvic bone, and soft tissue based on the linear‐exponential, plateau, full mechanistic, and specific mechanistic sarcoma dose‐response model. According to the models, all treatment plans resulted in similar risks of secondary bladder or rectal cancer and pelvic bone or soft tissue sarcoma except for the estimated risk of the bladder according to the full mechanistic model using IMRT(6MV;FF) technique compared with VMAT techniques with FFF options. The overall estimation of EAR indicated that the radiation‐induced cancer risk due to RT in HRPC was lower for bladder than the rectum. EAR values ranged from 1.47 to 5.82 for bladder and 6.36 to 7.94 for rectum, depending on the dose–response models used. The absolute risks of the secondary pelvic bone and soft tissue sarcoma were small for the plans examined. We theoretically predicted the radiation‐induced secondary cancer risk in HRPC patients with pelvic irradiation. Nevertheless, prospective clinical trials, with larger patient cohorts with a long‐term follow‐up, are needed to validate these model predictions.


| INTRODUCTION
Prostate cancer is the second most common malignancy in men in the world. 1 Radiotherapy (RT) plays an essential role in nearly every stage of the prostate cancer. According to the European Association of Urology Guidelines, intensity-modulated RT (IMRT), with long-term androgen deprivation therapy, is the main treatment option for high-risk prostate cancer (HRPC) patients. 2 Prophylactic pelvic nodal irradiation has failed to show beneficial results in cN0 patients. 3 The STAMPEDE trial time to failure-free survival was worse in patients with N+ disease (HR, 2.02 [95% confidence interval (CI), 1.46-2.81]) than in those with N0 disease. 4 Hence, the pelvic lymph node irradiation is still a confounding problem for clinicians, and it is important to carefully weigh the relative advantage of adding the pelvic lymph node irradiation over its potential toxicity or increased secondary cancer risk in the HRPC patients.
The development of secondary cancers after RT is an undesired outcome of the therapy that can be observed in long-term cancer survivors. There are several factors that impact secondary cancer risk which can be analysed from a radiobiological perspective: age at irradiation, type of irradiated tissue, irradiated volume, treatment technique, previous irradiation/radiological investigations. 5 In this study, secondary cancer risks were assessed using Schneider's concept of the organ equivalent dose (OED) and excess absolute risk (EAR). The EAR of developing secondary cancer after exposure to radiation can be estimated from the organ specific differential dose-volume histograms (dDVHs) based on biological models that are fitted to data from atomic bomb survivors and Hodgkin patients treated with RT. [6][7][8][9][10] To our knowledge, no previous report has estimated radiation-induced secondary cancer risks in HRPC patients with pelvic irradiation.
The study aimed to estimate radiation-induced secondary cancer risks using pelvic irradiation in HRPC patients in modern RT techniques with different energy levels and flattening filter (FF) options.

2.A. | Treatment planning volumes
Computed tomography data of five biopsy-proven HRPC patients was randomly selected from our database for a retrospective planning study. Two different planning target volumes (PTV 1 and PTV 2 ) were delineated for each patient. The PTV 1 included the prostate, seminal vesicles, and pelvic lymphatics, while PTV 2 included only the prostate and seminal vesicles (Fig. 1). The mean PTV 1 and PTV 2 volumes of five patients were 1022.30 and 278.94 cm 3 respectively.
The bladder, rectum, and femoral heads were considered organs at risk (OAR) (Fig. 1). In addition, all pelvic bones were contoured, and pelvic soft tissue was created as a structure (total scanned volume in the irradiation field minus all pelvic bones and PTVs).

2.B. | Treatment planning
IMRT and volumetric modulated arc therapy (VMAT) techniques were used to generate treatment plans. The prescribed dose was 54 Gy in 27 fractions for the PTV 1 with a sequential boost (24 Gy in 12 fractions for the PTV 2 ). At least 95% of the PTVs were required to receive ≥95% of the prescription dose. The dose-volume constraints used for the OARs are listed in Table 1. In total, five different treatment plans with 6 and 10 MV photon energies were created for each patient, including IMRT plans with FF and VMAT plans with FF and flattening filter-free (FFF) irradiation modes. Herein, the plans are referred to as IMRT (6MV;FF) , VMAT (6MV;FF) , VMAT (6MV;FFF) , VMAT (10MV;FF) , and VMAT (10MV;FFF) .
While the IMRT plans consisted of 11 equally spaced coplanar beams, the VMAT plans used three full coplanar arcs. All plans were optimized in the Eclipse treatment planning system (version13.6, Varian Medical Systems) using the identical optimization parameters for the targets and OARs. Once each plan was optimized, the dDVH's for the bladder, rectum, pelvic bone, and pelvic soft tissue were extracted and used to calculate the OEDs and EARs for each plan type and for all study cases.

2.C. | Calculation of second malignancy risk estimates
The risk of developing a secondary solid cancer after RT is usually represented by the EAR. The EAR to develop 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 persons per year (PY). 9 The EAR can be calculated according to Eq. (1): 1. An example of the contour of the planning target volumes (PTVs) and organs at risk (OARs) on the axial plane.
T A B L E 1 Dose constraints of the organs at risk (OARs).

OAR
Goal or constraint dose where EAR 0 is the initial slope of the dose-response curve at a low dose and the RED (risk equivalent dose) is the dose-response relationship for radiation-induced cancer in units of dose. The function μ takes into account the age of the population examined, based on the patient's age at the time of irradiation (agex), and the attained age of the patient in years (agea). The factor μ is given by Eq. (2): where γ e and γ a are age modifying factors (EAR 0 was originally calculated for persons exposed at age 30 where V T is the total organ volume. The sum is taken over all the bins of the dose-volume histogram, and V Di is the volume of the organ that is exposed to dose D i. It is known, that for doses below 2 Gy, the dose-response relation is linear. 11 For higher doses and inhomogeneous dose distributions, however, the dose-response relation is no longer linear 6,11 and other dose-response functions are required to describe their relation. To facilitate the estimation of secondary cancer risks in the irradiated organs, Schneider et al. 6,7 introduced the concept of the OED which assumes that any two dose distributions in an organ are equivalent if they cause the same radiation-induced cancer incidence. 6 The OED is then described by the following expression in Eq. (4): and the EAR org can be estimated as in Eq. (5): The OED values are independent of the EAR 0 and modifying function µ. 9 The EAR 0 values for Western populations and different sites were taken from the study by Schneider et al. 9 and tabulated in Table 2. There are different models for the RED calculation, based on different assumptions of cells' behavior after dose exposition. 7,9 The linear model assumes a linear response over the whole dose range: The mechanistic model that accounts for killing and fractionation effects: 8,9 where 0α ¼ α þ β D DT d T The parameters α and β are those from the linear quadratic model of cell killing, describing the linear and quadratic dose-response of the tissue to radiation. The parameter R describes the repopulation and ability to repair between dose fractions delivered.
For this model, there are two limiting cases with the parameter R = 0 for no repair and R = 1 for full repopulation/repair occurring. Therefore, the limit R = 0 leads to the linear-exponential model: The linear exponential model takes into account that the probability for cell killing increases exponentially with dose which leads to a decrease of the risk of cancer induction due to the killing of mutated cells.
Moreover, the limit R = 1 is the case of full repopulation/repair, known as the plateau model: Previous studies have demonstrated that, for inhomogeneous dose distributions above 4 Gy, the full mechanistic, plateau, and linear-exponential models represent a better description of the dose-response function than the linear model. 7,9 Therefore, the linear model was not included in our results.
OEDs for the bladder and rectum were calculated for the three different dose-response models based on the dDVHs according to: 1ÀR Di (10) The risk of radiation-induced pelvic bone and soft tissue secondary malignancies were calculated using a specific mechanistic sarcoma model based on an intermediate repopulation (R = 0.5) according to: The site-specific parameters were derived from a combined fit to data from atomic bomb survivors and Hodgkin patients treated with single doses of 2-40 Gy assuming an α/β value of 3 Gy. 9 The difference between the baseline risks of developing cancer, without exposure to radiation for the Japanese and Western populations was included. The parameters used for the EAR calculations are shown in  Epidemiological data from Japanese atomic bomb survivors and children exposed to radiation for medical reasons suggest that excess relative risks for radiation-induced cancers at a given attained age are substantially higher for individuals who are exposed during childhood than for those exposed at older ages. [20][21][22][23][24] However, the weight of epidemiological evidence now suggests that, for adult exposures, radiation-induced cancer risks do not generally decrease with increasing age at exposure. [25][26][27] In the Life Span Study cohort of the Japanese atomic bomb survivors, the excess relative risks for radiation-induced cancers as a function of age at exposure were examined. 27 As expected, the excess relative risk for cancer induction was higher during childhood and decreased progressively at exposure ages of [30][31][32][33][34][35][36][37][38][39][40]. Surprisingly, the excess relative risk of developing solid tumors ramped up again for exposure ages higher than 40 years old. 27 In addition, Richardson et al. observed that the radiation doses received after the age of 45 showed a stronger association with cancer mortality than those received at younger ages. 28 All these findings suggest that radiation sensitivity, measured in terms of carcinogenic events, increases with age among adults after the age of 40-45. Therefore, prostate cancer is a disease of relatively older patient but also it can be considered as a suitable group for the risk estimation of secondary cancer. The mean values and standard deviation of the EAR averaged over all patients. The bold value indicate a significantly lower EAR value compared with VMAT (6MV;FFF) and VMAT (10MV;FFF) . Abbreviations: IMRT, intensity-modulated radiotherapy; VMAT, volumetric modulated arc therapy; FF, flattening filter; FFF, flattening filter free; lin-exp, linear-exponential; plat plateau, mech full mechanistic, and sarcoma specific mechanistic sarcoma dose-response model. EAR unit is per 10 000 persons per year.
delivered lower doses to the OAR compared to pelvic irradiation plans due to the smaller volume of the PTV. As a result, the lower doses to the OARs contributed to the reduction of all calculated EARs with respect to the pelvic irradiation technique.
In the present study, we showed the radiation-induced secondary cancer risks of pelvic irradiation in all organs involved in the treatment of HRPC for a specific age group in this patient population according to different dose-response models. When comparing techniques, IMRT-PI (6MV;FF) plan significantly reduced the secondary cancer risk of the bladder according to the full mechanistic model compared with VMAT plans with FFF options, even though the absolute risks of secondary bladder cancer were small for the plans examined. This result may be attributed to the fact that the mechanistic model includes repair and repopulation effects. 8,9 As for the rectum, unlike the bladder, the secondary cancer risk in HRPC patients was independent of the treatment techniques, energies, modes, and dose-response models used. Additionally, this study showed that the absolute risks of the secondary pelvic bone and soft tissue sarcoma were small for the plans examined.
Regardless of the risk model selected, our findings revealed that VMAT did not increase the predicted risk when compared to IMRT, despite VMAT distributing lower dose over a larger volume of normal tissue than IMRT. 39,40 We showed that the FFF beam plans resulted in very small differences in EARs compared with the equivalent FF beam plans, for the bladder, rectum, pelvic bone, and soft tissue in our study. This finding agrees well with the results by Murray et al. 37 However, their results showed some statistically significant differences between the FFF and FF beams for calculated secondary cancer risks in the organs outside of the treatment fields.
There are a number of limitations in our work. We performed this analysis for the secondary cancer risks for only five patients selected from our patient data base. Other studies on similar cancer risk assessments also used a small number of patient cases, typically 2-3 cases per study. The reason for a relatively small sample size in this type of study is that the primary interest is to investigate the differences between the planning techniques rather than the factors due to interpatient variability. 30,32,[37][38][39][40][41][42] There are also other uncertainties in radiation-induced secondary cancer models and parameters. Schneider's concept of OED was employed as this incorporates the fractionation, and repair and repopulation (with the mechanistic model). 9 Models based on the full and no repair/repopulation, were also adopted to illustrate a range of possibilities for estimating the secondary cancer risk factors. The results of this study show secondary cancer risks for all the OARs from pelvic irradiation in HRPC with three different dose-response models. However, we did not assess the impact of hormonal therapy on secondary cancer risks for the HRPC patients, even though it is accepted that the systemic cancer treatment with chemotherapy and hormonal therapy is also associated with increased risk of secondary malignant neoplasm. 43 Another aspect of relevance for the cancer risks from modern RT techniques is the use of imageguidance techniques employing ionizing radiation. Risk evaluations accounting simultaneously for primary radiation and imaging doses have shown that the additional risk due to repeated imaging is in fact very small. 44 Such doses were not available for the patients used retrospectively for this study, and therefore we did not include the impact of image-guided radiotherapy on secondary cancer risk. Finally, we did not attempt to estimate the secondary cancer risks in out-offield organs. In contrast with our study which looked into only in-field F I G . 2. The excess absolute risk (EAR) of the bladder, rectum, pelvic bone and, soft tissue. The EAR based on differential dose-volume histograms of (a) the bladder, (b) rectum, (c) pelvic bone, and soft tissue for pelvic irradiation. Calculation of the EAR was performed using the linear-exponential (dark blue), plateau (blue), and full mechanistic (light blue) dose-response models for the bladder and rectum. The EAR calculation was performed using the sarcoma model for the pelvic bone (dark blue) and soft tissue (blue).
The mean values per 10 000 PY per Gy averaged over all five patients are shown. IMRT intensity-modulated radiotherapy, VMAT volumetric modulated arc therapy, FF flattening filter, FFF flattening filter free.
or close-to-the field organs and tissues, there appears to be a real difference between the plans using the FFF beams versus FF beams with the FFF beams resulting in plans with smaller calculated risk factors in out-of-field organs. [34][35][36][37] This is a reasonable and somewhat straightforward expectation since there is no flattening filter material in the beam path for a FFF beam, therefore the FFF beam produces less scatter in the beam production that results in less deposited dose to the organs outside of the direct treatment field. Despite this difference, the calculated absolute risk benefits from the FFF at large distances from the irradiated area, were very small for prostate cancer patients in the age range we considered in our study. Nevertheless, treatment of younger patients would have possibly resulted in greater absolute benefits from the FFF beams.
In our study, we attempted to minimize different planning system software and treatment margins since both may contribute to different secondary cancer risk estimates by creating plans, using the same planning system, and using the same dose constraint and PTV margin.

| CONCLUSION
A major strength of this study is its novelty; to our knowledge, this was the first study to predict the risk of secondary cancer incidence following RT in HRPC patients with pelvic irradiation. We compared radiation-induced secondary cancer risks using several contemporary,

CONF LICT OF I NTEREST
The authors declare that there is no conflict of interest regarding the publication of this article.