Since the implementation in HL therapy of extended field irradiation, the high cure rate was offset by late side effects and development of SMNs in a relevant fraction of patients . The progresses in imaging and the better knowledge of the disease biology, with consequent better prognostic stratification of patients, have allowed a decrease in the therapeutic load consisting in a progressive reduction of chemotherapy cycles, radiation dose and treated volume in most patients . However, the risk of late iatrogenic effects remains remarkable. The radiation delivery techniques can heavily condition the distribution of the dose in tissues and alter the toxicity profile of a treatment. Involved-field IMRT has shown excellent target coverage and amelioration of side effects in a clinical study by Lu et al. . Volumetric modulated arc therapy has been shown to significantly reduce hearth dose in HL patients affected with cardiovascular disease  and to perform better than IMRT in sparing the OARs when using involved nodal RT [10, 26]. With the same purpose Tomotherapy has been recently proposed for the treatment of HL . Recent preliminary studies of proton beam therapy for mediastinal HL have been reported . The OARs toxicities and development of second breast neoplasms would be expected to be reduced by the use of particle therapy. However, while it is epidemiologically reasonable to expect that a dose reduction is associated with a reduced risk of late effects, an improvement in SMN risk due to dose reduction is not yet clearly established.
Our study aims at analyzing 5 different radiation delivery techniques in three different hypothetical scenarios of supradiaphragmatic HL through a comprehensive dosimetric study. The main endpoint was to investigate, for each single technique, the balance between the predicted OARs injuries and the predicted development of SMNs, with the same target optimal dose coverage. The advantage of IMRT for heart and left ventricle sparing as well as its disadvantages in the low dose region, in particular for breasts, have been already reported in the literature . However, in our study the above advantages and disadvantages were quantified and extended to other state-of-the-art techniques.
As surrogate indicators of OARs morbidities, some of the constraints recently suggested by the literature were used. We chose constraints predictive of feared radioinduced injuries commonly described in patients treated with sequential chemo-radiotherapy for HL such as hypothyroidism , asymptomatic cardiac valvular dysfunction  and radiation pneumonitis  and specifically extrapolated from HL patients’ cohorts, together with some other more general constraints suggested by QUANTEC reviews .
Lacking epidemiological data relative to the recent RT delivery techniques, estimation of the risk of SMN for breasts, lungs, and thyroid based on mathematical models [19, 20] was used. Many uncertainties are involved in modelling the underlying biology of radiation induced-cancer. Nevertheless, these models may be reliably used to predict the impact on SMN induction of a given technique relative to another reference technique. To this end, we introduced the concept of risk ratio RR as a parameter for plan evaluation.
It should be noted that for the quality factor Q (stochastic RBE) we take a value of 1.1 for protons. There is a strong energy dependence for the quality factor and a factor of two, as recommended by ICRP 92 , would perhaps only be expected at very low energies in the tail of the Bragg peak. The main contribution to the normal tissue integral dose, however, will come from the plateau region of the Bragg curve due to the protons passing through normal tissue to reach the target volume. This portion of the Bragg curve consists predominantly of dose deposited by higher energy protons (much higher than 8 MeV) for which the NCRP quotes a value of one [32, 33]. In the normal tissue distal to the target volume, although the quality factor may be higher, the irradiated volume will be very much smaller and the deposited dose will be lower due to the finite maximum range of protons in the tissue. Therefore, it is safe to assume that the vast majority of normal tissues will be irradiated by protons with a quality factor close to one.
As regards PTV coverage, in the framework of a satisfactory performance of all the above techniques, the optimal coverage was obtained by TOMO and PRO plans.
As far as constraint compliance is concerned, in all PTV scenarios, AP-PA, FIMRT, and IMRT plans exceed the whole-heart-V25 of 10%. This limit, associated with a <1% probability of cardiac mortality, is an overly safe risk estimate based on model predictions and consequently the risk may be overestimated . The other constraints were met by all five techniques in PTV1 and PTV2 scenarios. For PTV3, the AP-PA, FIMRT and IMRT also failed to meet left atrium V25 cutoff volume of 63% which is a significant predictor of mitral and aortic valvular defects. The latter are particularly important for those patients characterized by high cure rates and prolonged survival like HL patients because of their progressive nature and potential contribution to overt cardiac toxicity . Only the AP-PA failed to meet thyroid V30 dose constraint predictive of hypothyroidism. Remarkably, beyond DVH predictors, TOMO and PRO led to a reduction in the doses to all the OARs compared with the other plans.
Conversely, the estimated risk ratio of SMNs induction for breasts and lungs was significantly increased by IMRT and TOMO in all scenarios though it is lower when the target volume is larger. No relevant risk ratio increase in thyroid cancer was found for any technique. To be noted, theoretically PRO led to a reduction of risk ratio in all cases. Among photon delivery techniques, conventional AP-PA and FIMRT resulted in the lowest estimated risk of SMNs.
This study, exploring the trade-offs between radio-induced toxicities and SMN by planning comparative evaluations, provides informative tools so as to evaluate which HL patient potentially deserves a more advanced radiation technique obtaining a real advantage in terms of deterministic and/or stochastic damage prevention. Diverse variables must be considered such as individual patients features, site and size of disease in order to establish strategies capable of performing a risk-adapted radiotherapy.
Let us point to some potential limitations of our proof-of-concept study. First, we considered one single model case not taking into account morphological differences peculiar to each single patient such as heart, lung and breast volumes. We also analyzed three different PTVs that, although paradigmatic, did not cover all possible varieties of HL. Moreover, in SMN estimation the uncertainty linked to neutron RBE for carcinogenesis should be taken into account .
Given the above considerations, our analysis suggests that, as already shown for other tumor sites [17, 34], proton therapy could theoretically be the optimal radiation modality in all HL scenarios studied, provided that plan robustness and organ motion are properly managed . However, costs and availability currently limit proton usage. Regarding photon techniques, the choice of the more appropriate treatment should be tailored to the individual case. For instance, for a young male patient with a large tumor or a patient with cardiac co-morbidity both requiring a total dose of 30 Gy, TOMO plan would result extremely advantageous. On the contrary, TOMO could not be equally advantageous for a good prognosis young (25 years) HL bearing female patient requiring a total dose of 20 Gy, which implies a very low risk of late organ injuries. In such a case, radioinduced breast cancer may be of more concern and FIMRT may result more appropriate.