Radiotherapy has been often described as a “two edged sword” because, if on one hand it is a major modality of cancer treatment, but on the other it can be a cause of cancer. During the past decade, radiation-induced secondary malignancies have become a major concern and recent studies have shown that radiotherapy treatment is associated with a small, though statistically significant, enhancement in the risk of secondary cancers. At present it is generally agreed that around 10% of patients may develop a second malignancy due to radiation therapy, even if this number is not known with much certainty and could range between 6% and 13% [1,2]. One of the largest study, carried out on about 29,000 patients which received radiotherapy after surgery of breast cancer, showed an increase of cancer risk in non-affected breast from 7.5% to 9.3% after 15 years [3]. Another study performed by Brenner et al. [4] in more than 120,000 patients with prostate cancer showed an increased cancer risk after radiotherapy of 6% if compared to the group that underwent only surgery.

Modern radiotherapy techniques are moving in the direction of optimizing the dose conformation to tumors meanwhile sparing the exposure of organs at risk and minimizing the radiation load to healthy tissues; this is usually obtained by improving patient positioning, target localization and providing sharp dose gradients. In this context, new high-precision technologies like intensity modulated radiotherapy (IMRT) and SBRT represent major advances in cancer treatment. The high degree of conformity associated with these techniques is often obtained by increasing the number of fields and using fixed-shape or dynamic conformal arc beams. This has important implications in the debate over the possible increase of secondary cancers due to radiation therapy, essentially for two reasons. Firstly, delivery of a specified dose from these special techniques requires the accelerator to be energized for longer (more monitor units are needed) and, as a consequence, the total body dose due to leakage radiation is increased by a factor of two or three [5]. Secondly, if on the one hand these techniques lead to tight dose conformations and sharp dose gradients, on the other they are likely to increase the integral dose exposure to non-tumor tissues since larger volumes of normal tissues can be exposed to lower doses [6].

The increase of energy deposition in healthy tissues might play a leading role in the induction of secondary cancers, especially in the light of past and recent literature data which show a possible correlation between integral dose and secondary malignancies [7,8,9,10].

The aim of this work was to compare the integral dose (ID) imparted by SBRT and 3D-CRT and establish possible correlations between integral dose calculated from differential DVHs and the increase of carcinogenic risk.

The ID attempts to describe energy deposition within the whole body and it is historically considered as a physical quantity capable of representing the “physical aggression” and risk of complications due to radiation therapy. Integral dose is the product of mass of tissue irradiated and absorbed dose. Although it is generally accepted that normal tissue complication risk and secondary malignancies risk increase as the ID increases, ID is rarely used in clinical practice to compare competing plans or to evaluate treatment outcome. At present, it is still unknown which increase of integral dose could be considered clinically acceptable; however, as a general rule, it is recommended to keep ID to a minimum, tumor dose being fixed and provided that normal tissues are not unacceptably compromised [11]. Over the last decade many studies have attempted to compare integral dose received by different x-ray irradiation techniques (e.g., IMRT vs. 3D-CRT). In these studies ID was calculated by DVHs or by the product of mean dose and irradiated volume [12,13,14,15,16]. However, in none of these studies was an attempt made to estimate the potential consequences of integral dose increase.

To our knowledge, there are no studies investigating ID in stereotactic body radiation therapy and comparing SBRT with “traditional” techniques. Generally, SBRT does not lead to unacceptable side effects if the serial organs are excluded from high dose regions and the organs at risk constraints are respected. Most importantly, we wondered whether an increase of ID may be related to an increase in secondary cancer risk. This issue is not trivial since ID calculation does not consider fractionation effects, which are supposed to play a key role in radiation-induced malignancies.

Furthermore, the induction of secondary cancers is a matter of interest in SBRT as the use of this technique might be extended to patients who live for many years after radiotherapy. Given the very encouraging results with the SBRT technique, phase III studies are in fact strongly needed in order to compare SBRT with surgery in operable patients [17]. Lastly, considering the dose-escalation and retreatment possibilities provided by SBRT, the overall integral dose might reach levels considerably higher than achieved with conventional planning techniques and delivery schemes.

As a result of the hypofractionated dose delivery scheme and the higher sensitivity to fractionation of late-responding tissues, in all SBRT plans the NTID increased compared to 3D-CRT plans (p = 0.002) (Table 3) while, as expected, no significant difference of ID to PTVs were observed between the techniques (Table 4), with p = 1. In fact, both for 3D-CRT and SBRT, the lesions were planned with the intent of providing the same target coverage and maintaining approximately the same tumor average dose.

For each patient, EARs for the main organs of interest were calculated (Figure 3). As a general rule, EARs for SBRT patients are smaller than secondary cancer risk for patients that received standard conformal radiation therapy. Exceptions are represented by patients 5 and 6, where EARs for left lung (patients 5 and 6) and esophagus (patient 6) are higher for SBRT rather than 3D-CRT.

Secondary cancer risk for all solid tumors was also calculated for each patient (Figure 4). Except for patient 6, EARs for SBRT are systematically lower than EARs obtained for 3D-CRT.

As a general rule, non tumor integral dose depends on a number of factors. As reported by D’Souza [13] beam margin size and beam energy are the most relevant parameters, with smaller margin and higher energy consistently reducing NTID regardless of the number of beams. In the present study, given the same beam energy (6 MV), one would expect that the smaller margin used in SBRT (2 mm vs. 5 mm used in 3D-CRT) lead to a reduction of the NTID. Actually, the hypofractionated dose delivery scheme used in SBRT increases the (normalized) average dose to non tumor tissues, thus increasing the NTID as well; in fact, according to the dose normalization rule, low α/β ratio are more sensitive to higher dose per fraction. According to the same work, number of beams, beam direction and relative beam weight have little effect on NTID.

When evaluating radiation dose to healthy tissues, especially at sites remote from the treatment region, radiation leakage may represent an important factor, increasing integral dose to normal structures. Specifically, two main sources of leakage can be considered to increase the patient NTID: transmission of radiation through the collimator leaves and leakage through the primary collimation system. It is generally believed that leakage through the leaves of a conventional collimator might carry a significant contribution to peripheral integral dose (leakage is about 2.5% for 6 MV photon beams).

More specifically, leakage radiation might play a crucial role in some special techniques. A recent work by Petti et al. [33] showed that peripheral dose in CyberKnife radiosurgery is due largely to leakage radiation. They measured the radiation dose at different distances outside the treatment field finding that for distances larger than 40 cm from the field edge radiation leakage is the dominant component and is directly related to the number of MU delivered. Dose values due to leakage radiation resulted two to five times higher than those measured for the comparable gamma knife brain treatment, and up to a factor of four times larger those measured in the IMRT experiment.

Over the last decade integral dose has aroused a lively interest due to its alleged correlation with secondary cancers. Radiation-induced secondary malignancies are rare, but since treatment techniques improve and clinical outcomes are improving accordingly, secondary tumor risk after oncologic treatment might represent a relevant issue. Of course, the risk of secondary cancer induction from radiation treatments is likely not to be worrisome within a few years after treatment given the latency period of malignancies, but radiation-induced secondary cancer might be a relevant concern for those (especially young) patients whose progression free survival is greater than 5 years. Different works show that disease-free survival is rapidly increasing for patients which undergo SBRT; Uematsu and et al. [34] reported 5-year results of treatment of 50 patients with stage I NSCL finding a 3-year survival rate being as high as 66%, while Nagata et colleagues [35] found overall survival rates after 3 years for stage IA and IB lung cancer being 83% and 72%, respectively.

In the present paper we used a dose-response relationship for cancer induction that includes fractionation effects, therefore suitable for radiotherapy applications. Under the assumptions made and according to the model used our study indicates that despite the fact that for all patients integral dose is higher for SBRT treatments than 3D-CRT, secondary cancer risk associated to SBRT patients is significantly smaller than that calculated for 3D-CRT. This suggests that integral dose may not be a good estimator for quantifying cancer induction. This is imputable to the fact that in the case of SBRT the dose-response curve for carcinoma induction is highly non-linear. In the present study we assumed a bell-shaped behavior, consequently leading to lower cancer induction rates. Furthermore, integral dose does not consider fractionation at all, and fractionation is supposed to be directly correlated to an increased cancer risk [31]. Our findings are in agreement with a recent study by Schneider and colleagues [31] showing that hypofractionated radiotherapy has the potential for secondary cancer reduction.

It is worth noting that the results presented in this study are valid under the assumptions made: neither neutron dose nor leakage radiation accounted and risk calculations performed on the plane of interest for the treatment. In the present analysis a limitation of DVH computation might be represented by the use of Pencil Beam algorithm, which is known to have some drawbacks in low density media. Nevertheless, possible dosimetric inaccuracies are likely to affect DVH calculations (and the following DVH-related evaluations) in both TPS in the same magnitude. Despite the use of more advanced dose evaluations algorithm is encouraged for lung treatments (anisotropic analytical algorithm and collapsed cone convolution), pencil beam algorithm is still widely used and implemented in the clinical practice both for standard 3D-CRT and SBRT of lung malignancies [36,37]. In such case, voltages below 6 MV are always recommended and dose evaluations with heterogeneity corrections are necessary [38,39].

Finally, some consideration about the application of the linear-quadratic (LQ) model are needed. In fact, in radiotherapeutic applications, the LQ formalism is the tool most commonly used for quantitative predictions of dose/fractionations dependencies. Questions may arise when using the LQ model to describe dose response in the high dose per fraction range.

At present, the LQ model is reasonably well validated (experimentally and theoretically) and its use is reasonable up to several Grays per fraction. In addition, there is a fairly wide range of studies for which it is possible to test concordance with the LQ predictions in the 2 to 20 Gy range [35]. Different works show that different quantitative in vivo endpoints are consistent with LQ model over a wide range of doses per fraction, including those of interest in hypofractionation [40].

At higher doses the LQ model might have deficiencies and experimental survival curves suggest a purely linear rather than a continuation of the linear-quadratic shape (continuously bending) so LQ model may not properly evaluate the tumor dose, predicting more cell kill.

Caution is advised in the presence of bio-mathematical model which use radiobiological parameters. In fact, although mathematical models are widely used to compare different radiotherapy technique and different fractionation scheme, the radiobiological parameters on which they are funded are still not completely optimized, thus unavoidably introducing some degree of approximation.

Our results indicate that although NTID was greater in SBRT than in 3D-CRT, secondary cancer risk associated to SBRT patients is significantly smaller than that calculated for 3D-CRT. This suggests that integral dose may not be a good estimator for quantifying cancer induction.

The development of reliable secondary cancer risk models seems to be a key issue in fractionated radiotherapy and comparisons of integral dose received with 3D-CRT and other special techniques are strongly encouraged.