Abstract
Background/Aim: Carbon ion radiotherapy (CIRT), proton therapy (PT) and intensity-modulated radiotherapy (IMRT) are new radiation modalities suitable for treatment of spinal sarcomas. The objective of the study was to compare the treatment planning of these modalities. Patients and Methods: We conducted a treatment planning comparison of the three modalities using a phantom imitating a spinal sarcoma and then six actual cases with spinal tumors. A uniform biological effective dose (BED) of 90 Gy10 was prescribed in previously reported fractionation schedules for each modality. The surface/center spinal cord dose constraints were set to BED of 96/77 Gy(E)3, respectively. Results: CIRT achieved better homogeneity of dose distribution and coverage of target than PT independently of tumor extent around the spinal cord. In IMRT plans, the spinal cord dose was higher than that under CIRT and PT and coverage of the target deteriorated depending on the tumor extension. Conclusion: CIRT was most appropriate for the treatment of advanced spinal sarcomas.
- Carbon ion radiotherapy
- proton therapy
- patch-field technique
- intensity-modulated radiotherapy
- spinal sarcoma
Radiotherapy is utilized for treatment of unresectable spinal sarcomas, although it is necessary to deliver higher doses, exceeding the tolerance dose of spinal cord, to the target in order to eradicate the disease (1-3). Intensity-modulated radiotherapy (IMRT) has been reported as a preoperative and postoperative radiotherapy with favorable local control, but radical radiotherapy with IMRT has rarely been reported (4). Carbon ion radiotherapy (CIRT) and proton therapy (PT) are significantly different approaches characterized by better conformity to the target and less adverse effects on the organs-at-risk (OARs). In particular, the patch-field technique has been developed to irradiate a target encircling a critical organ, which is applicable to CIRT and PT by using their steep penumbra. Although PT using the patch-field technique has been reported for several diseases (5, 6), there are few investigations regarding the use of the patch-field technique with CIRT (7). Even though the outcome of the three modalities should be evaluated by a prospective clinical trial, it would be difficult to carry-out such a trial because of the rarity of the disease and small number of Institutions dealing with all three modalities.
In the present study, a treatment planning comparison of CIRT, PT with the patch-field technique and IMRT was conducted for spinal sarcomas surrounding the spinal cord. We initially evaluated physical characteristics of these modalities using a virtual phantom imitating a spinal sarcoma and verified the differences in the treatment planning for six actual cases.
Patients and Methods
Configuration of the phantom and patient selection. A columnar virtual phantom with a 50-mm diameter imitating a spinal sarcoma was generated (Figure 1). The phantom was located in the center of a 30-cm cube as a substitute for the patient's body. Computed tomography (CT) value of 10 HU was set for the whole structure, including the phantom and the external cube. The phantom was hollowed out centrally to represent the spinal canal containing the spinal cord, with a margin of 5 mm. CIRT, PT and IMRT plans were generated for this phantom.
We retrospectively selected six patients with lumbar spinal metastases of carcinoma, assuming spinal sarcomas, surrounding the spinal cord to various degrees. The target volume ranged from 30.2 to 120.1 ml (median 85.6 ml), and all of the tumors contacted the spinal cord. In two patients, the tumor encircled half the spinal cord; in another two patients, the tumor encircled three-quarters of the spinal cord, and in the remaining two patients, the tumor encircled the whole circumference of the spinal cord.
Treatment planning. For the planning comparison in the actual cases, we used CT images taken with 3-mm slice thickness in the supine position. The tumor extent was evaluated by CT, magnetic resonance imaging (MRI) and positron-emission tomography (PET), if available. The clinical target volume (CTV) included the gross tumor volume plus a 5-mm margin for potential microscopic invasion. The planning target volume (PTV) included the CTV plus a 5-mm safety margin for positioning errors. Where the tumor was located close to the spinal cord, the margins were reduced accordingly. A safety margin for the spinal cord was not added.
We used Eclipse software version 10.0 (Varian Medical Systems, Palo Alto, CA, USA) for delineation of the targets and OARs and for the treatment planning for PT and IMRT. In the PT plans, passively scattered proton beams with accelerated energy of 250 MeV/n were used and the dose calculation algorithm was the proton convolution superposition (8, 9). In the IMRT plans, 10-MV photon beams were used, and the dose calculation algorithm was the anisotropic analytical algorithm (10). For the treatment planning of CIRT, the delineation data of the targets and OARs were converted to the DICOM-RT format and transferred to XiO-C software ver. 4.47 (Elekta, Stockholm, Sweden). In the CIRT plans, passively scattered carbon ion beams with accelerated energies of 290 MeV/n and 400 MeV/n were used and the dose calculation algorithm was pencil beam (8).
A uniform biological effective dose (BED) was set for the prescribed dose among the three modalities. Dose-fractionation schedule was determined based on previously reported treatment approaches for spinal sarcomas, that is, 77.4 Gy in 43 fractions for the IMRT (4), 77.4 Gy equivalents [GyE; physical dose in Gy × relative biological effectiveness (RBE)] in 43 fractions for the PT (1, 2, 11), and 64 GyE in 16 fractions for the CIRT (7, 12, 13). All of the dosages were equivalent to a BED of 90 Gy10. In the same way, the spinal cord surface/center doses were constrained to 63/54 GyE or Gy in 43 fractions for the PT and IMRT (1, 4, 11, 14) and 48/41 GyE in 16 fractions for the CIRT based on previous reports. These doses were equivalent to a BED of 96/77 Gy3 or GyE3. In IMRT, PT and CIRT, RBE value was set to 1.0, 1.1 and 3.0, respectively (15, 16).
The treatment plans for IMRT were performed applying volumetric modulated arc therapy (VMAT) with single arc irradiation. The collimator was rotated to 30° to minimize the contribution of the tongue-and-groove effect. In the optimization, the highest priority was given to the surface and center spinal cord doses being within the constraint level, whereas the PTV was irradiated with the prescribed dose homogeneously as much as possible. The VMAT plans were conducted using a high-definition multileaf collimator consisting of 60 pairs of leaves, which included 40 pairs of 5-mm central leaves and 20 pairs of 10-mm peripheral leaves.
The treatment plans for the CIRT and PT were performed using the patch-field technique (5, 7). The target region was divided into two volumes, and the divided regions were irradiated by respective beams from different directions, sparing the critical organs. By using a sharp dose fall-off after the Bragg peak, the lateral field edge of one field (the main field) was matched with the distal edge of the second field (the subfield). Both beams were created preferentially so as to comply with the spinal cord tolerance dose, while delivering the prescribed dose to PTV secondarily. Patch-fields were created as the combination of two pairs of evenly weighted beams for each tumor in order to offset hot/cold spots at the combined area (the patch region). Each irradiation field was shaped using the 5-mm multileaf collimator for CIRT and a patient collimator made by cutting a brass block for PT.
Overview of the virtual phantom imitating a spinal sarcoma. A column-shaped phantom (50 mm in diameter) imitating a spinal tumor was created. The central hole (20 mm in diameter) represents the spinal canal, which contains the spinal cord (10 mm in diameter). The margin between the tumor and the spinal cord was set at 5 mm. This phantom was set in the center of the 30-cm cube with the computed tomography value of 10 HU as a substitute for a patient's body.
Treatment plan evaluation. Dose–volume histograms (DVHs) for the targets and OARs for each plan were computed. The evaluated DVH parameters included the mean doses irradiated to the PTV (Dmean), the doses received by at least 95%, 90% and 80% of the PTV (D95%, D90%, D80%, respectively), the homogeneity index [HI; (maximum dose received by the PTV)÷(minimum dose received by the PTV)] and the maximum dose irradiated to the surface and center of the spinal cord (Dsurface and Dcenter, respectively). These parameters were compared using unpaired Welch's t-tests to assess the differences among the three modalities. The conventional value of p<0.05 was considered significant in those assessments.
Three-dimensional dose-distribution in the phantom study. Plans for carbon ion radiotherapy (CIRT) (a-c), proton therapy (PT) (d-f), intensity-modulated radiotherapy (IMRT) (g-i). The linearity of the distal end of subfields was considerably different between CIRT and PT plans (arrowheads).
Results
Phantom study. The dose distributions in the phantom are shown in Figure 2. The most remarkable difference between the CIRT and PT plans was the linearity of the distal end of the subfield. In the PT subfield, bulged edges of the dose distribution were prominent at the distal end of the irradiated area (Figure 2e, f). In contrast, the CIRT plan showed the linear linkage of the two fields (Figure 2b, c). Another point to note was the lateral diffusion of the proton beams (Figure 2d). In the PT plan, the main field spread laterally more and more as the beam progressed, which encompassed a mismatch of the patch region and the main field edge. These phenomena resulted in more favorable coverage in the CIRT plan as shown in Figure 3.
In the IMRT plan for the phantom, the target was covered homogeneously but the dose gradient around the spinal cord was shallower compared to the CIRT and PT plans. Consequently, the central dose to the spinal cord was considerably high (Figure 2g, h and Figure 3).
Case study. Comparison of DVH parameters of the three modalities. Figure 4 shows the dose distributions for CIRT, PT and IMRT for a representative patient with tumor encircling the whole circumference of the spinal cord. The distributions of the DVH parameters of the six cases are summarized in Figure 5.
The spinal cord doses were approximately within the constraint levels and were considered to be clinically acceptable in all three modalities' plans. However, in all of the IMRT plans, the central doses (83.7-100.0% of the constraint) as well as the peripheral doses (92.7-101.6% of the constraint) to the spinal cord exhibited near-constraint levels, while in the CIRT and PT plans the central dose (30.2-73.2% of the constraint for CIRT, 33.0-81.3% of the constraint for PT) was much lower than the peripheral dose (94.4-100.4% of the constraint for CIRT, 81.6-99.4% of the constraint for PT) and would not exceed the constraint levels.
Comparing coverage of PTV, CIRT achieved more favorable dose distributions than PT (Dmean; p=0.047, D90%; p=0.009, respectively) (Figure 5). In IMRT plans, there was great variability of the coverage among the cases.
Case study. Correlation between extension around spinal cord and DVH parameters. The CIRT, PT and IMRT plans provided remarkably different results regarding the correlation between tumor coverage and the tumor extension around the spinal cord (Figure 6). In CIRT plans, there was no definite correlation between the coverage and tumor extension around the spinal cord, while PT plans demonstrated slightly worsening trend of the coverage depending on the tumor extension. Furthermore, in IMRT plans, the coverage deteriorated considerably, particularly in the cases with tumor encircling the whole circumference of the spinal cord.
Discussion
In recent years, several newly emerging radiation modalities have been introduced and have led to a transformation of the role of radiotherapy in the treatment of bone and soft-tissue tumors. IMRT has contributed towards the achievement of a more conformal dose to the tumor and a lower dose to the spinal cord, and has been proven to be effective in the treatment of spinal metastases, though the improvement of local control for radioresistant tumors has remained limited (4).
Favorable treatment outcomes of particle therapies for bone and soft-tissue sarcomas have been reported (1, 2, 7, 12, 17). In particular, CIRT was advocated to have a high local control rate for spinal sarcomas (7, 17). In particle therapy for tumors surrounding critical organs, it is advantageous to patch two independently delivered dose distributions in order to treat a single target volume. This patch-field technique has been most commonly used in passively scattered PT for head and neck or spinal tumors to optimize the dose distribution within a target volume in close proximity to critical normal structures (5, 18). Although each dose distribution would be combined with the other, the total dose distribution at the patch region would usually be heterogeneous because the gradients of the two dose distributions are not identical.
Dose-volume histograms of the phantom study.
Patch-field plans in passively scattered PT may not be acceptable in a certain number of cases because the dose heterogeneity is produced by combining the dose distributions from two separate fields, even though some investigators have added refinements to the treatment planning. Hug et al. reported that they used three to four combinations of fields of anterior/anterior oblique and lateral/lateral oblique field directions with varying beam obliquities and match lines on alternating days to eliminate potential dose heterogeneities (5). Li et al. recently reported a novel patch-field technique using an optimized grid filter for passively scattered proton beams (19).
In the present study, the phantom study and the actual case studies showed consistent results. As mentioned above, IMRT achieved the most homogeneous dose distribution. However, the target dose homogeneity and coverage in the IMRT plans deteriorated with increasing tumor extension around the spinal cord, which indicates that IMRT is unsuitable for advanced spinal tumors compared to CIRT and PT. In addition, the central spinal cord dose was relatively high due to the shallower dose gradient around the spinal cord. Although several investigators adopted spinal cord dose constraints similar to those used in the present study and reported no spinal cord injuries, the tolerance level of the spinal cord has not been established, especially when the superficial portion of the cord is exposed to high-dose radiation (1, 20, 21). Therefore, the integral dose, as well as the maximum dose irradiated to the spinal cord, should be reduced as much as possible to prevent radiation-induced myelitis.
Dose distribution of a representative case with tumor encircling the whole circumference of the spinal cord. a: carbon ion radiotherapy (CIRT), b: proton therapy (PT), c: intensity-modulated radiotherapy (IMRT).
Box plots of the parameters of the dose–volume histogram among the six cases. The top and bottom of each box represent the third and first quartiles, respectively. The middle line indicates the median. a: Dsurface and b: Dcenter, the maximum dose irradiated to the surface/center of the spinal cord; c: homogeneity index (HI); d: Dmean, the mean dose irradiated to the PTV; e: D95%, f: D90%, g: D80%, Dx% means a dose received by at least x% of the PTV. *Significant difference.
Correlation between extension around spinal cord and dose–volume histogram parameters. CIRT: carbon ion radiotherapy, PT: proton therapy, IMRT: intensity-modulated radiotherapy. a: Homogeneity index (HI); b: Dmean, the mean dose irradiated to the PTV; c: D95%; d: D90%, Dx% means a dose received by at least x% of the PTV.
Carbon ion beams are known for their smaller lateral penumbra and the sharper dose fall-off after the Bragg peak than proton beams. Weber et al. reported that the broadening of a proton beam is approximately 3.5 times larger than that of a carbon beam when compared for the same depth range (22). For this reason, the lateral edge of the main field in PT plans would stray more from the junctional line of the main field and subfield compared to CIRT plans. Weber et al. also reported that a spike of dose distribution was formed behind the target volume when proton and carbon beams were passing the edge of a dense structure, such as bone, and that this integral dose of the spikes was much larger for protons than for carbon ions (22). This phenomenon consequently results in an irregular distal end of the subfield and inconsistence with the junctional line, since the subfield usually passes through a vertebral bone. These irregularities of the main field and subfield result in a mismatch with each other, leading to hot and cold spots at the patch region more significantly in PT plans. Certainly, a carbon ion beam creates a dose tail at the distal side of the Bragg peak, which possibly encompasses the irregularity of the distal end. However, it consists of a low linear energy transfer component and is considered to have less biological effect on OARs. In summary, CIRT is theoretically considered to be more appropriate for the patch-field technique than PT.
To our knowledge, this is the first report comparing the dose distribution between CIRT, PT using the patch-field technique, and IMRT for patients with bone tumors. Although physical properties of carbon ion beams relative to proton beams were well-known, it has not been clarified what effect the difference would have in clinical use, especially in treatment for spinal tumors. The results of this study revealed that the irregularity of the dose distribution in proton beams reduced the quality of the patch-field technique compared with carbon ion beams. It is true that the dose distribution could be produced differently depending on the oncologist because the patch-field technique is strongly influenced by the optimization process. However, our findings indicate that the best efforts would fail to compensate for the shortcoming of the beam profile. We also found that the IMRT plans, which do not require the patch-field technique, resulted in more homogeneous dose distributions compared to the particle therapies but resulted in insufficient coverage for the target area adjacent to the spinal cord, particularly for the tumors that completely encircled the circumference of the spinal cord, which might lead to marginal recurrences. Nonetheless, the spinal cord doses were higher than those of the particle therapies.
Recent years, scanning ion beam radiotherapy has been introduced. It is a new irradiation method for particle therapies and is expected to be applied to spinal sarcomas in the near future. For the comparison to existing radiation methods, further investigation is needed.
In conclusion, the results of this planning comparison study indicate the efficacy of CIRT for the treatment of advanced spinal tumors. Because it corroborates the reported clinical outcome, CIRT is considered to be an encouraging alternative for the treatment of unresectable spinal tumors.
Acknowledgements
This work was supported in part by Grant-in-Aid for Scientific Research, Japan (grant no. 26670563 and 26461894).
- Received April 11, 2015.
- Revision received May 12, 2015.
- Accepted May 14, 2015.
- Copyright© 2015 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved
