- Email: [email protected]
Tumor control probability and the utility of 4D vs 3D dose calculations for stereotactic body radiotherapy for lung cancer
Medical Dosimetry 40 (2015) 64–69
Medical Dosimetry journal homepage: www.meddos.org
Tumor control probability and the utility of 4D vs 3D dose calculations for stereotactic body radiotherapy for lung cancer Gilmer Valdes, Ph.D.,* Clifford Robinson, M.D.,† Percy Lee, M.D.,‡ Delphine Morel ,§,║ Daniel Low, Ph.D.,‡ Keisuke S. Iwamoto, Ph.D.,‡ and James M. Lamb, Ph.D.‡ Department of Radiation Oncology, Perelman Center for Advanced Medicine, University of Pennsylvania, Philadelphia, PA; †Department of Radiation Oncology, Siteman Cancer Center, Washington University in St. Louis, St. Louis, MO; ‡Department of Radiation Oncology, David Geffen School of Medicine, UCLA, Los Angeles, CA; §Department of Biomedical Engineering, AIX Marseille 2 University, Marseille, France; and ║Department of Medical Physics, Joseph Fourier University, Grenoble, France
*
A R T I C L E I N F O
A B S T R A C T
Article history: Received 16 December 2013 Received in revised form 12 September 2014 Accepted 5 October 2014
Four-dimensional (4D) dose calculations for lung cancer radiotherapy have been technically feasible for a number of years but have not become standard clinical practice. The purpose of this study was to determine if clinically significant differences in tumor control probability (TCP) exist between 3D and 4D dose calculations so as to inform the decision whether 4D dose calculations should be used routinely for treatment planning. Radiotherapy plans for Stage I-II lung cancer were created for 8 patients. Clinically acceptable treatment plans were created with dose calculated on the end-exhale 4D computed tomography (CT) phase using a Monte Carlo algorithm. Dose was then projected onto the remaining 9 phases of 4D-CT using the Monte Carlo algorithm and accumulated onto the end-exhale phase using commercially available deformable registration software. The resulting dose-volume histograms (DVH) of the gross tumor volume (GTV), planning tumor volume (PTV), and PTVsetup were compared according to target coverage and dose. The PTVsetup was defined as a volume including the GTV and a margin for setup uncertainties but not for respiratory motion. TCPs resulting from these DVHs were estimated using a wide range of alphas, betas, and tumor cell densities. Differences of up to 5 Gy were observed between 3D and 4D calculations for a PTV with highly irregular shape. When the TCP was calculated using the resulting DVHs for fractionation schedules typically used in stereotactic body radiation therapy (SBRT), the TCP differed at most by 5% between 4D and 3D cases, and in most cases, it was by less than 1%. We conclude that 4D dose calculations are not necessary for most cases treated with SBRT, but they might be valuable for irregularly shaped target volumes. If 4D calculations are used, 4D DVHs should be evaluated on volumes that include margin for setup uncertainty but not respiratory motion. Copyright & 2015 American Association of Medical Dosimetrists.
Keywords: 4D dose calculations 3D dose calculations TCP
Introduction Great efforts have been made to try to incorporate information about respiratory motion into the calculation of the dose received by tumors and critical normal structures in lung cancer radiotherapy treatments.1-3 With the development of respiratorycorrelated 4-dimensional computed tomography (4D-CT), it is now possible to acquire 4D data representative of patients' true lung motion. A 4D-CT data set generally consists of 8 to 10 3D-CT
This work is supported in part by NIH, USA R01CA096679 and R01CA116712. Reprint requests to: Gilmer Valdes, Ph.D., Department of Radiation Oncology, University of Pennsylvania, 3400 Civic Center, Philadelphia, PA 19104-6021. E-mail: [email protected]. http://dx.doi.org/10.1016/j.meddos.2014.10.002 0958-3947/Copyright Ó 2015 American Association of Medical Dosimetrists
data sets, with each data set corresponding to a specified phase of the respiratory cycle.4,5 A common way to characterize 4D-CT data is through the use of a maximum intensity projection (MIP) image to represent the entire motion envelop of tumor or normal structure throughout the breathing cycle, For tumor volumes, an internal target volume (ITV)6 may be generated using the MIP image as a guide. Planning is often accomplished on a separate free-breathing scan or a representative phase (typically end inhale or end exhale) of the breathing cycle from the 4D-CT data. As such, plan evaluation occurs using a static 3D representation of the dose. Whether this 3D estimation is an accurate representation of the dose that the gross tumor volume (GTV) and critical normal structures would receive because of radiation delivery during respiration is still debated.1–3,7-10 It is now possible to calculate
G. Valdes et al. / Medical Dosimetry 40 (2015) 64–69
4D dose distributions by using software that implements methods of deformable image registration. In general, it is thought that this 4D dose distribution will serve as a better estimate of the real dose that different structures would receive.9,10 Several prior publications have compared 4D dose distributions to 3D dose distributions to investigate the effect of tumor motion on the calculation of the dose received by the different structures in the patient.1-3,7,8 All of these authors found some differences between the 4D and 3D dose distributions. Nevertheless, a controversy remains on whether the difference between 3D and 4D calculations is significant. Rao et al.1 reported only minor differences between 3D and 4D calculations; Starkschall et al.3 reported that for 26% of their patients, the 4D calculations indicated a difference in target volume dose sufficiently great to warrant replanning. If we take into account that 4D dose calculations may take up to 10 times longer than 3D calculations, bringing 4D calculations to the clinical routine might prove difficult. Therefore, the goal of this work was to determine whether 4D dose calculations were significantly different from 3D dose calculations, and if so, would such differences practically affect tumor control probability (TCP) such that 4D doses need to be calculated for highly mobile tumors. The present work builds on previously published reports in that we consider not only the planning target volume (PTV) and the GTV as target structures but also a volume we call the PTVsetup, which is an expansion of the GTV with margins sized to take into account setup uncertainty but not tumor motion. Furthermore, TCP models are used to estimate the clinical relevance of observed dosimetric differences.
65
based on the MIP image derived from the entire 4D-CT data set and checked against the end-exhale and end-inhale 4D-CT phases. A GTV was calculated and contoured separately on the end-exhale phase for analysis purposes but not used explicitly in planning. A PTV was created by expanding the ITV by 6 mm in the superoinferior direction and 4 mm in the anteroposterior and left-right directions to account for daily setup variations and breathing motion uncertainty beyond that accounted for in the ITV. Additionally, a separate volume (denoted PTVsetup) was created by expanding the GTV (end exhale) by 4 mm in all directions, as a representation of setup error alone because motion was accounted for by the 4D dose calculation (Fig. 1). In a static situation, there is no ambiguity in referring to this volume as the PTV, but under respiratory tumor motion, there is ambiguity in the definition of the PTV owing to the fact that it includes a margin explicitly for respiratory motion and for statistical setup uncertainty, which is independent of respiration. For that reason, we defined the PTVsetup as the GTV plus a 4 mm expansion, to account for setup uncertainty unrelated to respiration. The value of 4 mm for the expansion follows from our clinical experience for setup uncertainty in the context of stereotactic body radiation therapy (SBRT) not subject to respiratory motion. The PTVsetup is not part of the clinical routine and was not used for dose planning, it was only for interpretation of the 4D dose. The targeted plan normalization was that 95% of the PTV should be covered by the prescribed dose. For peripheral tumors (7 patients) defined by tumors being outside of a 2-cm radius around the proximal bronchial tree volume, the treatment was planned for 3 fractions of 18 Gy. For central tumors (1 patient), the treatment was planned for 4 fractions of 12.5 Gy. The same optimization objectives and penalties that are currently used in our clinic were used for these mock plans. Conformality was constrained by thresholds of VRx/VPTV, V50%/VPTV, and Dmax at 2 cm from the PTV, per Radiation Therapy Oncology Group protocol 0618. Fixed-field intensity-modulated radiation therapy (IMRT) plans were generated for Varian Novalis Tx using 6MV photons. Normal structure dose constraints from Radiation Therapy Oncology Group 0618 were also used. The ribs were required to receive no greater than the prescription dose, and the PTV was sometimes trimmed off the ribs to achieve that goal if it was deemed safe to do so. Plan optimization and Monte Carlo dose calculation was carried out using iPlan version 4.1, with grid spacing of 2 mm and a variance of 1% as settings. The dose was optimized on the end-exhale 4D-CT phase 3D image. These then served as the reference plan for subsequent 4D dose calculations (Table 2).
Methods and Materials 4D-CT image acquisition
4D dose calculations
Overall, 8 patients with American Joint Committee on Cancer seventh edition T1 to T2 lung tumors were imaged with 4D-CT as part of an institutional review board–approved research protocol. Patients were selected for participation in the protocol because they had lower lobe lung tumors that were expected to exhibit substantial breathing motion. The tumor position, volume, and motion extent for all patients is shown in Table 1. Tumor motion extent (excursion) was determined using the 3D displacement of the center of mass of the tumor between end-exhale and end-inhale phases. All patients were imaged using 4D-CT with a Siemens Biograph 40 positron emission tomography/CT (Siemens AG, Erlangen, Germany), with breathing wave recorded by an abdominal bellows (Philips Healthcare, Andover, MA). Each 4D-CT image set was composed of 10 amplitude-sorted phases spaced from the fifth to the 95th percentiles of breathing amplitude in steps corresponding to 10 percentile points. Because the phases were chosen at fixed percentiles rather than fixed amplitude spacing, each phase corresponded closely to the configuration taken by the lung during approximately 10% of the breathing cycle. No specific breathing control (abdominal compression, breath hold, etc.) was employed during the scans. Treatment planning Data obtained from 4D-CT imaging of the 8 patients was imported into the iPlan 4.1 (Brainlab AG, Feldkirchen, Germany) treatment planning system, and tumor volumes and normal structures were contoured as per standard clinical practice at our institution. Critical normal structures (ipsilateral and contralateral lungs, spinal cord, heart, esophagus, bronchial tree, ribs, trachea, and the GTV) were delineated on the reference end-exhale 4D-CT scan. An ITV was contoured Table 1 GTV position, volume and excursion Patient
Position
Excursion (mm)
GTV (cm3)
1 2 3 4 5 6 7 8
Right lower lobe Left lower lobe Right lower lobe Right lower lobe Left upper lobe Left lower lobe Left lower lobe Right lower lobe
7.8 38.8 16.7 30.8 3.4 9.7 12.9 29.5
20.3 7.7 4.5 6.3 18.1 13.6 3.7 16.9
To calculate the 4D dose, the reference plan created earlier using the endexhale phase was applied to all different phases for each patient using iPlan Monte Carlo with the same settings as in the 3D calculations. As a result, 3D doses were obtained for each phase. After this, the plans were exported in DICOM-RT format to MIM 5.5 (MIM Software, Cleveland, OH), where all phases were registered to the end-exhale phase using the deformable image registration algorithm provided by MIM. For every patient, the deformable registration was validated by visually inspecting the registration of landmarks near the tumor, and the edges of the tumor itself. In 3 patients where the tumor movement was greater than 2 cm, the registration algorithm was manually assisted by a rigid prealigment of the tumor. The 4D dose distribution for each patient was then calculated, assuming that the prescribed monitor units in an IMRT plan were equally distributed over 10 breathing phases. As noted in the Methods and Materials, the 4D-CT phases were explicitly reconstructed at amplitudes corresponding to uniform steps of 10 percentile points of breathing amplitude between 5% and 95%. Finally, DVHs from both 3D and 4D calculations were obtained. Plans were evaluated by the minimum dose received by 99% of the GTV and the PTVsetup as well as the mean doses to those volumes. A paired Student t-test comparing the values of these dosimetric indices corresponding to DVHs from 3D and 4D calculations was performed.
TCP calculations The TCP of inhomogenous 3D and 4D dose distributions for different fractionation schedules (18 Gy 3, 12.5 Gy 4, 15 Gy 3, and 2 Gy 33) were calculated. The first 3 fractionated regimes are representative of SBRT plans, whereas the last one is representative of a commonly used fractionated regime.11,12 In the present study, a model similar to those used by different authors to calculate TCP for inhomogenous dose distributions was used.13-15 For purposes of simplicity, the reference plans were all created with doses prescribed per the primary SBRT regimens detailed in section Treatment Planning and subsequently scaled to the other aforementioned regimens for the sole purpose of TCP calculation without further modification to the plan itself. Given the variability that can be found in the literature for suggested alpha parameters and tumor cell density, the TCPs were calculated for 2 different sets of alpha and beta parameters: α ¼ 0.19 ⫾ 0.02 Gy1, β ¼ 0.02 ⫾ 0.002 Gy2, and α ¼ 0.30 ⫾ 0.02 Gy1, β ¼ 0.02 ⫾ 0.002 Gy2 and 2 different tumor cell density: 107 cells/cm3, 108 cells/cm3. These sets of parameters represent the lower and upper bounds commonly found in the literature.12,16 A more detailed description of how the TCPs were calculated is included in the Appendix.
66
G. Valdes et al. / Medical Dosimetry 40 (2015) 64–69
Fig. 1. GTV (magenta contour), PTVsetup (red contour), and PTV (green contour) for patient 2. (Color version of figure is available online.)
Results The values of several dosimetric indices that are regularly used in our clinic to compare treatment plans and evaluate DVHs are shown for both 3D and 4D calculations in Tables 3 and 4. The results of a paired Student t-test comparing the values of these dosimetric indices corresponding to 3D and 4D DVHs are also shown in those tables. The dosimetric indices corresponding to the PTV show statistically significant differences when 3D and 4D DVHs are compared. For all patients, PTV dosimetric indices were lower in the 4D calculations than in the 3D calculations, as expected. Conversely, concerning the GTV and the PTVsetup, differences between the results of the 4D and the 3D calculations were minor, except in patient 7. The 4D calculations for patient 7 rendered an undercoverage of the PTVsetup when compared with the 3D calculations, which rose to a level that would be concerning for clinical effect (D994D ¼ 46.9 Gy and D993D ¼ 52.1 Gy, a difference of
5.2 Gy). The shape of this tumor was extremely irregular, and as such, the resulting ITV and PTV were also very irregular, as can be seen in Fig. 2. Excluding patient 7, the largest deficit of the D994D with respect to D993D in either the GTV or the PTVsetup was 1.8 Gy. Finally, using 3D and 4D DVHs corresponding to the GTV for each of the patients, the TCPs for 4 different dose schedules, 2 different sets of alpha and beta parameters, and 2 different tumor cell densities were calculated. As can be seen from Tables 5 and 6, the TCPs for the schedules representative of SBRT (18 Gy 3, 12.5 Gy 4, and 15 Gy 3) are almost always near enough to one for all different parameters such that at these dose levels and the TCP dose response curve are in the saturation zone for all patients, including patient 7. In this region, changes in the dose result in small changes in the TCP, as can be seen when the total dose is changed from 54 Gy to 45 Gy in Tables 5 and 6. For the schedule representative of a conventionally fractionated regimen, the TCP is nearly 0 for both tumor cell densities and the first set of parameters (α ¼ 0.19 Gy1 ⫾ 0.02 Gy1; β ¼ 0.02 Gy2 ⫾ 0.002 Gy2).
Table 2 Definitions of dosimetric indices used to approve and compare treatment plans Quantity
Region of interest
Description
Vp Vp Vp D95 D99 (GTV) D99 (PTVsetup) MD (GTV) MD (PTVsetup) Dmin Dmin Dmax Dmax
GTV PTVsetup PTV PTV GTV PTVsetup GTV PTVsetup GTV PTVsetup GTV PTVsetup
Fractional volume of GTV receiving at least the prescribed dose Fractional volume of PTVsetup receiving at least the prescribed dose Fractional volume of PTV receiving at least the prescribed dose Dose received by at least 95% of the PTV Dose received by at least 99% of the GTV Dose received by at least 99% of the PTVsetup Mean dose received by the GTV Mean dose received by the PTVsetup Minimum dose received by the GTV Minimum dose received by the PTVsetup Maximum dose received by the GTV Maximum dose received by the PTVsetup
MD ¼ mean dose.
G. Valdes et al. / Medical Dosimetry 40 (2015) 64–69
67
Table 3 Dosimetric indices used for plan approval and comparison Vp (GTV)
Patient
Vp (PTVsetup)
Vp (PTV)
D95 (PTV)
D99 (GTV)
D99 (PTVsetup)
3D
4D
3D
4D
3D
3D10
4D
3D
3D10
4D
3D
4D
3D
4D
1 2 3 4 5 6 7 8
98.3 100.0 100.0 100.0 100.0 100.0 98.4 100.0
97.9 100.0 99.9 100.0 100.0 100.0 78.6 100.0
92.5 97.9 96.0 94.5 100.0 97.7 89.2 91.0
89.8 99.4 96.2 96.9 95.7 98.6 56.9 97.6
95.3 96.0 96.7 94.0 95.1 93.6 89.1 91.8
95.7 95.7 96.1 95.6 79.2 84.1 81.3 97.3
70.2 58.7 78.4 72.6 81.0 74.8 38.6 75.1
54.1 54.3 55.0 53.3 50.1 53.4 53.7 52.8
54.4 54.4 54.5 54.4 48.4 50.9 52.5 55.6
46.1 26.3 43.7 37.7 48.4 38.2 40.8 38.9
52.8 55.6 59.0 56.8 53.1 56.1 53.9 56.6
52.3 56.8 59.4 61.6 51.7 56.5 52.5 58.7
48.3 53.3 43.3 49.5 50.9 53 52.1 50.1
47.2 54.4 44.7 51.2 49.1 53.7 46.9 52.5
Mean Paired t-test
99.6
97.0
94.8
91.4
93.9
90.6 S
68.7
53.3
53.1 S
40.0
55.5
56.2
50.1
NS
NS
50.0
NS
NS
3D10 is the 3D dose calculated using phase 10. NS ¼ not significant; S ¼ significant. Table 4 Dosimetric indices used for plan approval and comparison MD (GTV)
MD (PTVsetup)
Dmin (GTV)
Dmax (GTV)
Dmin (PTVsetup)
Dmax (PTVsetup)
Dmin (PTV)
3D
4D
3D
4D
3D
4D
3D
4D
3D
4D
3D
4D
3D
3D10
4D
3D
3D10
4D
60.8 57.4 62.9 66.7 55.2 60.3 55.8 60.2
58.7 58.3 62.7 67.4 54.0 59.1 54.6 62.3
59.4 56.7 60.9 64.0 53.9 58.7 55.3 58.2
57.5 57.4 60.8 62.7 52.6 57.8 54.0 60.5
49.8 55.3 53.3 54.1 52.3 56.1 53.4 53.6
48.7 56.3 53.5 59.0 50.9 55.6 51.9 55.1
63.7 59.2 65.6 72.1 57.2 62.8 57.7 62.9
60.9 59.7 65.2 69.9 55.6 61.5 56.4 64.9
39.3 50.7 34.7 39.6 49.6 48.9 48.4 45.8
37.7 49.7 34.5 44.1 47.5 52.1 43.6 45.3
64.1 59.3 65.6 72.1 57.3 62.8 59.2 63.9
61.0 59.8 65.2 69.9 55.6 61.4 57.5 64.9
41.4 45.9 36.8 41.3 45.5 42.3 49.2 43.0
43.1 44.8 36.6 41.1 44.1 40.5 47.9 45.6
31.9 14.3 28.8 23.2 44.4 24.5 25.3 19.5
64.1 60.4 65.6 75.4 57.3 63.8 60.2 65.8
65.3 68.0 66.9 79.8 56.3 62.9 58.8 69.2
60.9 59.7 65.2 69.9 55.6 61.5 57.5 65.0
59.6
58.4
57.9
53.5
53.9
62.7
61.8
44.6
44.3
63.0
61.9
43.2
43.0 S
26.5
64.1
65.9 S
61.9
59.9 NS
NS
NS
NS
NS
S
Dmax (PTV)
MD ¼ mean dose; NS ¼ not significant; S ¼ significant.
Fig. 2. Irregularly shaped PTV of patient 7. The PTV is shown by the red contour and 99% isodose line according to the 4D calculation is shown by the green contour. (Color version of figure is available online.)
68
G. Valdes et al. / Medical Dosimetry 40 (2015) 64–69
Table 5 TCP calculated for α ¼ 0.19 ⫾ 0.02, β ¼ 0.02 ⫾ 0.002, 4 different dose schedules and 2 different cell densities Patients
Tumor cell density (cells/cm3)
TCP_54 Gy_3 3D
7
1 2 3 4 5* 6 7 8
Mean
n
4D
10 108 107 108 107 108 107 108 107 108 107 108 107 108 107 108
4 4 4 4 4 4 4 4 4
107 108
4 0.99 4 0.99
4 4 4 4 4 4
TCP_45 Gy_3
0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.96 0.99 0.99 0.99 0.99 0.99 0.99
4 4 4 4 4 4 4
4 4 4 4 4 4
3D 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.91 0.99 0.99 0.99 0.99 0.99 0.99
4 0.99 4 0.99
4 4 4 4 4
4 4 4 4
TCP_66 Gy_33 4D
0.99 0.91 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.95 0.99 0.99 0.99 0.98 0.99 0.95
4 0.99 0.97
4 4 4 4 4 4
4 4 4 4
3D 0.98 0.86 0.99 0.99 0.99 0.99 0.99 0.99 0.98 0.91 0.99 0.99 0.99 0.97 0.99 0.98
4 0.99 0.96
o o o o o o o o o o o o o o o o
4D 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01
o 0.01 o 0.01
o o o o o o o o o o o o o o o o
0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01
o 0.01 o 0.01
Patient 5 was prescribed 50 Gy in 4 fractions instead of 54 Gy in 3 fractions as it is customary in our clinic for central lung tumors.
The TCP has mean values of 0.73 and 0.70 for 3D and 4D DVHs, respectively, when the second set of parameters (α ¼ 0.30 Gy1 ⫾ 0.02 Gy1; β ¼ 0.02 Gy2 ⫾ 0.002 Gy2) is used. However, no statistically significant differences were found between TCPs calculated using 3D or 4D DVHs for SBRT or regular fractionated regimes. It should be noted that, for some tumors, the TCP is larger under the 4D calculations. That may be because of the effect of respiratory motion smoothing over relatively cold areas in these tumors.
the PTV. However, the results of these calculations on the PTV are an indication of the dose received by the tissue under the PTV in the reference CT image where it was contoured, rather than the dose received by the physical area where the tumor moves and which is targeted in the plan. Therefore, when 4D calculations are performed, uniform coverage of the accumulated 4D dose on the tissue region defined by the PTV in the reference CT image should not be sought, as it would result in an unnecessary dose to the patient. Conversely, evaluating 4D dose using only the GTV is not correct either because the GTV does not contain a margin for setup variations that are expected to occur according to a statistical probability distribution. For that reason, we introduced the concept of the PTVsetup for evaluation of 4D accumulated dose. The PTVsetup is an expansion of the GTV that includes margin for setup uncertainty but not for tumor motion. We chose a margin of 4 mm, based on our institution's experience with cone-beam CT image guidance for soft tissue targets in the context of SBRT.
Discussion We found significant differences in D95(PTV) when 3D dose calculations were compared with 4D dose calculations for 8 Stage I-II lung cancer radiotherapy plans. These results are similar to those found by Starkschall et al.3; however, our interpretation of them is different. The 4D calculations indicate an undercoverage of
Table 6 TCP calculated for α ¼ 0.30 ⫾ 0.02, β ¼ 0.02 ⫾ 0.002, 4 different dose schedules and 2 different cell densities Patients
Tumor cell density (cells/cm3)
TCP_54 Gy_3 3D
1 2 3 4 5* 6 7 8 Mean
n
TCP_45 Gy_3 4D
107 108 107 108 107 108 107 108 107 108 107 108 107 108 107 108
o o o o o o o o o o o o o o o o
107 108
o 0.99 o 0.99
0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99
o o o o o o o o o o o o o o o o
3D 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99
o 0.99 o 0.99
o o o o o o o o o o o o o o o o
TCP_66 Gy_33 4D
0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99
o 0.99 o 0.99
3D
4D
0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99
0.66 0.19 0.72 0.25 0.96 0.77 0.97 0.80 0.69 0.24 0.82 0.40 0.68 0.21 0.74 0.30
0.53 0.11 0.78 0.34 0.96 0.76 0.99 0.92 0.56 0.15 0.75 0.30 0.58 0.13 0.86 0.48
o 0.99 o 0.99
0.73 0.37
0.70 0.38
o o o o o o o o o o o o o o o o
Patient 5 was prescribed 50 Gy in 4 fractions instead of 54 Gy in 3 fractions as it is customary in our clinic for central lung tumors.
G. Valdes et al. / Medical Dosimetry 40 (2015) 64–69
Despite the dosimetric differences related to the PTV, when the 3D and 4D dose calculations corresponding to the GTV or the PTVsetup were analyzed, only 1 patient showed a dosimetric difference that rose to a level of clinical significance. We regarded a dose deficit of 5% or greater of the prescription dose as one that would be regarded as clinically significant in most clinics. In the case of patient 7, a 9% deficit in D99(PTVsetup) in the 4D vs the 3D calculations occurred. We hypothesize that such differences were seen because of to the irregular shape of the tumor and PTV (Fig. 2). No other dosimetric deficit was larger than 4% of the prescription dose. On the contrary, when TCP was calculated for these tumors using a wide range of alpha, beta, and cell density parameters taken from the literature, TCP for SBRT fractionations used in our clinic (18 Gy 3 for peripheral tumors and 12.5 Gy 4 for central tumors), TCP was almost always greater than 99% and always greater than 90%. For these SBRT fractionations, large changes in the total dose received by the GTV have negligible effect on the TCP because the TCP response curve is in the saturation zone. These analytical results are consistent with the published reports of lung SBRT that show that, regardless of tumor location or tumor motion, when adequate dose is applied and good image guidance is employed, local control has been in excess of 90%.17,18 To add more complexity to the discussion, it is important to highlight that interplay effects between the IMRT leaf movement and the respiration cycle were not included in our 4D dose calculation method. According to Rao et al.,1 calculating the 4D dose as described in this study offers similar results as taking into account the interplay between the IMRT leaf movement and the respiration cycle. This statement is equivalent to that made by Bortfeld et al.19 who showed that the reference dose for the case of intrafraction motion should be the average dose obtained over the path of the motion. However, as has been shown by Seco et al., these statements are only valid in general for those plans where the delivery time of each of the IMRT segments is an order of magnitude bigger than the breathing period (3 to 5 seconds). If the delivery time of each of the IMRT segments is on the same order of magnitude of the breathing period, the interplay effect could cause nonnegligible effects.20 It is important to highlight that the results presented here were obtained for a sample of patients with small and highly mobile tumors. The mean tumor excursion of our 8 patients was 21 mm, with a standard deviation of 11 mm, while the mean and RMS tumor volumes were 12 ⫾ 8.8 cm3. In comparison, Rao et al.1 studied 10 patients with a mean tumor excursion of 14 ⫾ 6.6 mm and a mean volume of 42 ⫾ 49 cm3, whereas Starkschall et al.3 studied 15 patients with a mean excursion of 7.0 ⫾ 6.0 mm and a mean volume of 107 ⫾ 109 cm3. We expect that on average, small tumors will show a larger difference between 3D and 4D calculations, although individual cases may vary substantially in their sensitivities to 4D calculations. A small tumor with large motion corresponds to a PTV whose volume is a large multiple of the GTV, at least relatively, when compared with a large tumor with small motion. Conventionally, the plan is normalized so that the prescription dose covers 95% of the PTV. The portion of the PTV not receiving the prescription dose is normally in the boundary region of the PTV. This boundary region of the PTV is proportionally larger compared with the GTV in a small tumor that moves to a large degree, compared to a large and relatively stationary tumor.
Conclusions Summarizing, it is likely that for lung stereotactic body radiotherapy with doses of 18 Gy 3 fractions prescribed to 95% of the
69
PTV, 3D calculations are adequate for treatment optimization as the DVHs estimated using this method produce similar TCP as those obtained using 4D calculations. However, if the resources are available, 4D calculations should be carried out especially for highly mobile tumors with irregular shapes. Furthermore, if 4D calculations are performed, the dose received by the PTVsetup, i.e., the GTV with a margin to account for setup uncertainty but not respiratory motion, should be evaluated in addition to evaluating dose to the GTV and the PTV.
Supplementary Materials Supplementary material cited in this article is available online at http://dx.doi.org/10.1016/j.meddos.2014.10.002.
References 1. Rao, M.; Wu, J.; Cao, D.; et al. Dosimetric impact of breathing motion in lung stereotactic body radiotherapy treatment using image-modulated radiotherapy and volumetric modulated arc therapy. Int. J. Radiat. Oncol. Biol. Phys. 83: e251–e256; 2012. 2. Rietzel, E.; Chen, G.T.; Choi, N.C.; et al. Four-dimensional image-based treatment planning: target volume segmentation and dose calculation in the presence of respiratory motion. Int. J. Radiat. Oncol. Biol. Phys. 61:1535–50; 2005. 3. Starkschall, G.; Britton, K.; McAleer, M.F.; et al. Potential dosimetric benefits of four-dimensional radiation treatment planning. Int. J. Radiat. Oncol. Biol. Phys. 73:1560–5; 2009. 4. Nagata, Y.; Wulf, J.; Lax, I.; et al. Stereotactic radiotherapy of primary lung cancer and other targets: results of consultant meeting of the International Atomic Energy Agency. Int. J. Radiat. Oncol. Biol. Phys. 79:660–9; 2011. 5. Low, D.A.; Nystrom, M.; Kalinin, E.; et al. A method for the reconstruction of four-dimensional synchronized CT scans acquired during free breathing. Med. Phys. 30:1254–63; 2003. 6. Hodapp, N. The ICRU Report 83: prescribing, recording and reporting photonbeam intensity-modulated radiation therapy (IMRT)]. Strahlenther. Onkol. 188:97–9; 2012. 7. Flampouri, S.; Jiang, S.B.; Sharp, G.C.; et al. Estimation of the delivered patient dose in lung IMRT treatment based on deformable registration of 4D-CT data and Monte Carlo simulations. Phys. Med. Biol. 51:2763–79; 2006. 8. Britton, K.R.; Starkschall, G.; Liu, H.; et al. Consequences of anatomic changes and respiratory motion on radiation dose distributions in conformal radiotherapy for locally advanced non–small-cell lung cancer. Int. J. Radiat. Oncol. Biol. Phys. 73:94–102; 2009. 9. Brock, K.K.; McShan, D.L.; Ten Haken, R.K.; et al. Inclusion of organ deformation in dose calculations. Med. Phys. 30:290–5; 2003. 10. Wang, H.; Dong, L.; O’Daniel, J.; et al. Validation of an accelerated ‘demons’ algorithm for deformable image registration in radiation therapy. Phys. Med. Biol. 50:2887–905; 2005. 11. Chi, A.; Liao, Z.; Nguyen, N.P.; et al. Systemic review of the patterns of failure following stereotactic body radiation therapy in early-stage non–small-cell lung cancer: clinical implications. Radiother. Oncol. 94:1–11; 2010. 12. El Sharouni, S.Y.; Kal, H.B.; Battermann, J.J. Tumour control probability of stage III inoperable non–small cell lung tumours after sequential chemoradiotherapy. Anticancer Res. 25:4655–61; 2005. 13. Niemierko, A.; Goitein, M. Implementation of a model for estimating tumor control probability for an inhomogeneously irradiated tumor. Radiother. Oncol. 29:140–7; 1993. 14. Tome, W.A.; Fowler, J.F. On cold spots in tumor subvolumes. Med. Phys. 29:1590–8; 2002. 15. Webb, S.; Nahum, A.E. A model for calculating tumour control probability in radiotherapy including the effects of inhomogeneous distributions of dose and clonogenic cell density. Phys. Med. Biol. 38:653–66; 1993. 16. Qi, X.; Low, D.A.; Kupelian, P.; et al. Analysis of outcomes in early stage non small cell lung cancer irradiation: fractionation scheme implications for stage I versus stage II disease. Int. J. Radiat. Oncol. Biol. Phys. 84(3):S566; 2012. 17. Timmerman, R.; Bastasch, M.; Saha, D.; et al. Stereotactic body radiation therapy: normal tissue and tumor control effects with large dose per fraction. Front. Radiat. Ther. Oncol. 43:382–94; 2011. 18. van Baardwijk, A.; Tomé, W.A.; van Elmpt, W.; et al. Is high-dose stereotactic body radiotherapy (SBRT) for stage I non–small cell lung cancer (NSCLC) overkill? A systematic review Radiother. Oncol. 105:145–9; 2012. 19. Bortfeld, T.; Jokivarsi, K.; Goitein, M.; et al. Effects of intra-fraction motion on IMRT dose delivery: statistical analysis and simulation. Phys. Med. Biol. 47:2203–20; 2002. 20. Seco, J.; Sharp, G.C.; Turcotte, J.; et al. Effects of organ motion on IMRT treatments with segments of few monitor units. Med. Phys. 34:923–34; 2007.
Medical Dosimetry journal homepage: www.meddos.org
Tumor control probability and the utility of 4D vs 3D dose calculations for stereotactic body radiotherapy for lung cancer Gilmer Valdes, Ph.D.,* Clifford Robinson, M.D.,† Percy Lee, M.D.,‡ Delphine Morel ,§,║ Daniel Low, Ph.D.,‡ Keisuke S. Iwamoto, Ph.D.,‡ and James M. Lamb, Ph.D.‡ Department of Radiation Oncology, Perelman Center for Advanced Medicine, University of Pennsylvania, Philadelphia, PA; †Department of Radiation Oncology, Siteman Cancer Center, Washington University in St. Louis, St. Louis, MO; ‡Department of Radiation Oncology, David Geffen School of Medicine, UCLA, Los Angeles, CA; §Department of Biomedical Engineering, AIX Marseille 2 University, Marseille, France; and ║Department of Medical Physics, Joseph Fourier University, Grenoble, France
*
A R T I C L E I N F O
A B S T R A C T
Article history: Received 16 December 2013 Received in revised form 12 September 2014 Accepted 5 October 2014
Four-dimensional (4D) dose calculations for lung cancer radiotherapy have been technically feasible for a number of years but have not become standard clinical practice. The purpose of this study was to determine if clinically significant differences in tumor control probability (TCP) exist between 3D and 4D dose calculations so as to inform the decision whether 4D dose calculations should be used routinely for treatment planning. Radiotherapy plans for Stage I-II lung cancer were created for 8 patients. Clinically acceptable treatment plans were created with dose calculated on the end-exhale 4D computed tomography (CT) phase using a Monte Carlo algorithm. Dose was then projected onto the remaining 9 phases of 4D-CT using the Monte Carlo algorithm and accumulated onto the end-exhale phase using commercially available deformable registration software. The resulting dose-volume histograms (DVH) of the gross tumor volume (GTV), planning tumor volume (PTV), and PTVsetup were compared according to target coverage and dose. The PTVsetup was defined as a volume including the GTV and a margin for setup uncertainties but not for respiratory motion. TCPs resulting from these DVHs were estimated using a wide range of alphas, betas, and tumor cell densities. Differences of up to 5 Gy were observed between 3D and 4D calculations for a PTV with highly irregular shape. When the TCP was calculated using the resulting DVHs for fractionation schedules typically used in stereotactic body radiation therapy (SBRT), the TCP differed at most by 5% between 4D and 3D cases, and in most cases, it was by less than 1%. We conclude that 4D dose calculations are not necessary for most cases treated with SBRT, but they might be valuable for irregularly shaped target volumes. If 4D calculations are used, 4D DVHs should be evaluated on volumes that include margin for setup uncertainty but not respiratory motion. Copyright & 2015 American Association of Medical Dosimetrists.
Keywords: 4D dose calculations 3D dose calculations TCP
Introduction Great efforts have been made to try to incorporate information about respiratory motion into the calculation of the dose received by tumors and critical normal structures in lung cancer radiotherapy treatments.1-3 With the development of respiratorycorrelated 4-dimensional computed tomography (4D-CT), it is now possible to acquire 4D data representative of patients' true lung motion. A 4D-CT data set generally consists of 8 to 10 3D-CT
This work is supported in part by NIH, USA R01CA096679 and R01CA116712. Reprint requests to: Gilmer Valdes, Ph.D., Department of Radiation Oncology, University of Pennsylvania, 3400 Civic Center, Philadelphia, PA 19104-6021. E-mail: [email protected]. http://dx.doi.org/10.1016/j.meddos.2014.10.002 0958-3947/Copyright Ó 2015 American Association of Medical Dosimetrists
data sets, with each data set corresponding to a specified phase of the respiratory cycle.4,5 A common way to characterize 4D-CT data is through the use of a maximum intensity projection (MIP) image to represent the entire motion envelop of tumor or normal structure throughout the breathing cycle, For tumor volumes, an internal target volume (ITV)6 may be generated using the MIP image as a guide. Planning is often accomplished on a separate free-breathing scan or a representative phase (typically end inhale or end exhale) of the breathing cycle from the 4D-CT data. As such, plan evaluation occurs using a static 3D representation of the dose. Whether this 3D estimation is an accurate representation of the dose that the gross tumor volume (GTV) and critical normal structures would receive because of radiation delivery during respiration is still debated.1–3,7-10 It is now possible to calculate
G. Valdes et al. / Medical Dosimetry 40 (2015) 64–69
4D dose distributions by using software that implements methods of deformable image registration. In general, it is thought that this 4D dose distribution will serve as a better estimate of the real dose that different structures would receive.9,10 Several prior publications have compared 4D dose distributions to 3D dose distributions to investigate the effect of tumor motion on the calculation of the dose received by the different structures in the patient.1-3,7,8 All of these authors found some differences between the 4D and 3D dose distributions. Nevertheless, a controversy remains on whether the difference between 3D and 4D calculations is significant. Rao et al.1 reported only minor differences between 3D and 4D calculations; Starkschall et al.3 reported that for 26% of their patients, the 4D calculations indicated a difference in target volume dose sufficiently great to warrant replanning. If we take into account that 4D dose calculations may take up to 10 times longer than 3D calculations, bringing 4D calculations to the clinical routine might prove difficult. Therefore, the goal of this work was to determine whether 4D dose calculations were significantly different from 3D dose calculations, and if so, would such differences practically affect tumor control probability (TCP) such that 4D doses need to be calculated for highly mobile tumors. The present work builds on previously published reports in that we consider not only the planning target volume (PTV) and the GTV as target structures but also a volume we call the PTVsetup, which is an expansion of the GTV with margins sized to take into account setup uncertainty but not tumor motion. Furthermore, TCP models are used to estimate the clinical relevance of observed dosimetric differences.
65
based on the MIP image derived from the entire 4D-CT data set and checked against the end-exhale and end-inhale 4D-CT phases. A GTV was calculated and contoured separately on the end-exhale phase for analysis purposes but not used explicitly in planning. A PTV was created by expanding the ITV by 6 mm in the superoinferior direction and 4 mm in the anteroposterior and left-right directions to account for daily setup variations and breathing motion uncertainty beyond that accounted for in the ITV. Additionally, a separate volume (denoted PTVsetup) was created by expanding the GTV (end exhale) by 4 mm in all directions, as a representation of setup error alone because motion was accounted for by the 4D dose calculation (Fig. 1). In a static situation, there is no ambiguity in referring to this volume as the PTV, but under respiratory tumor motion, there is ambiguity in the definition of the PTV owing to the fact that it includes a margin explicitly for respiratory motion and for statistical setup uncertainty, which is independent of respiration. For that reason, we defined the PTVsetup as the GTV plus a 4 mm expansion, to account for setup uncertainty unrelated to respiration. The value of 4 mm for the expansion follows from our clinical experience for setup uncertainty in the context of stereotactic body radiation therapy (SBRT) not subject to respiratory motion. The PTVsetup is not part of the clinical routine and was not used for dose planning, it was only for interpretation of the 4D dose. The targeted plan normalization was that 95% of the PTV should be covered by the prescribed dose. For peripheral tumors (7 patients) defined by tumors being outside of a 2-cm radius around the proximal bronchial tree volume, the treatment was planned for 3 fractions of 18 Gy. For central tumors (1 patient), the treatment was planned for 4 fractions of 12.5 Gy. The same optimization objectives and penalties that are currently used in our clinic were used for these mock plans. Conformality was constrained by thresholds of VRx/VPTV, V50%/VPTV, and Dmax at 2 cm from the PTV, per Radiation Therapy Oncology Group protocol 0618. Fixed-field intensity-modulated radiation therapy (IMRT) plans were generated for Varian Novalis Tx using 6MV photons. Normal structure dose constraints from Radiation Therapy Oncology Group 0618 were also used. The ribs were required to receive no greater than the prescription dose, and the PTV was sometimes trimmed off the ribs to achieve that goal if it was deemed safe to do so. Plan optimization and Monte Carlo dose calculation was carried out using iPlan version 4.1, with grid spacing of 2 mm and a variance of 1% as settings. The dose was optimized on the end-exhale 4D-CT phase 3D image. These then served as the reference plan for subsequent 4D dose calculations (Table 2).
Methods and Materials 4D-CT image acquisition
4D dose calculations
Overall, 8 patients with American Joint Committee on Cancer seventh edition T1 to T2 lung tumors were imaged with 4D-CT as part of an institutional review board–approved research protocol. Patients were selected for participation in the protocol because they had lower lobe lung tumors that were expected to exhibit substantial breathing motion. The tumor position, volume, and motion extent for all patients is shown in Table 1. Tumor motion extent (excursion) was determined using the 3D displacement of the center of mass of the tumor between end-exhale and end-inhale phases. All patients were imaged using 4D-CT with a Siemens Biograph 40 positron emission tomography/CT (Siemens AG, Erlangen, Germany), with breathing wave recorded by an abdominal bellows (Philips Healthcare, Andover, MA). Each 4D-CT image set was composed of 10 amplitude-sorted phases spaced from the fifth to the 95th percentiles of breathing amplitude in steps corresponding to 10 percentile points. Because the phases were chosen at fixed percentiles rather than fixed amplitude spacing, each phase corresponded closely to the configuration taken by the lung during approximately 10% of the breathing cycle. No specific breathing control (abdominal compression, breath hold, etc.) was employed during the scans. Treatment planning Data obtained from 4D-CT imaging of the 8 patients was imported into the iPlan 4.1 (Brainlab AG, Feldkirchen, Germany) treatment planning system, and tumor volumes and normal structures were contoured as per standard clinical practice at our institution. Critical normal structures (ipsilateral and contralateral lungs, spinal cord, heart, esophagus, bronchial tree, ribs, trachea, and the GTV) were delineated on the reference end-exhale 4D-CT scan. An ITV was contoured Table 1 GTV position, volume and excursion Patient
Position
Excursion (mm)
GTV (cm3)
1 2 3 4 5 6 7 8
Right lower lobe Left lower lobe Right lower lobe Right lower lobe Left upper lobe Left lower lobe Left lower lobe Right lower lobe
7.8 38.8 16.7 30.8 3.4 9.7 12.9 29.5
20.3 7.7 4.5 6.3 18.1 13.6 3.7 16.9
To calculate the 4D dose, the reference plan created earlier using the endexhale phase was applied to all different phases for each patient using iPlan Monte Carlo with the same settings as in the 3D calculations. As a result, 3D doses were obtained for each phase. After this, the plans were exported in DICOM-RT format to MIM 5.5 (MIM Software, Cleveland, OH), where all phases were registered to the end-exhale phase using the deformable image registration algorithm provided by MIM. For every patient, the deformable registration was validated by visually inspecting the registration of landmarks near the tumor, and the edges of the tumor itself. In 3 patients where the tumor movement was greater than 2 cm, the registration algorithm was manually assisted by a rigid prealigment of the tumor. The 4D dose distribution for each patient was then calculated, assuming that the prescribed monitor units in an IMRT plan were equally distributed over 10 breathing phases. As noted in the Methods and Materials, the 4D-CT phases were explicitly reconstructed at amplitudes corresponding to uniform steps of 10 percentile points of breathing amplitude between 5% and 95%. Finally, DVHs from both 3D and 4D calculations were obtained. Plans were evaluated by the minimum dose received by 99% of the GTV and the PTVsetup as well as the mean doses to those volumes. A paired Student t-test comparing the values of these dosimetric indices corresponding to DVHs from 3D and 4D calculations was performed.
TCP calculations The TCP of inhomogenous 3D and 4D dose distributions for different fractionation schedules (18 Gy 3, 12.5 Gy 4, 15 Gy 3, and 2 Gy 33) were calculated. The first 3 fractionated regimes are representative of SBRT plans, whereas the last one is representative of a commonly used fractionated regime.11,12 In the present study, a model similar to those used by different authors to calculate TCP for inhomogenous dose distributions was used.13-15 For purposes of simplicity, the reference plans were all created with doses prescribed per the primary SBRT regimens detailed in section Treatment Planning and subsequently scaled to the other aforementioned regimens for the sole purpose of TCP calculation without further modification to the plan itself. Given the variability that can be found in the literature for suggested alpha parameters and tumor cell density, the TCPs were calculated for 2 different sets of alpha and beta parameters: α ¼ 0.19 ⫾ 0.02 Gy1, β ¼ 0.02 ⫾ 0.002 Gy2, and α ¼ 0.30 ⫾ 0.02 Gy1, β ¼ 0.02 ⫾ 0.002 Gy2 and 2 different tumor cell density: 107 cells/cm3, 108 cells/cm3. These sets of parameters represent the lower and upper bounds commonly found in the literature.12,16 A more detailed description of how the TCPs were calculated is included in the Appendix.
66
G. Valdes et al. / Medical Dosimetry 40 (2015) 64–69
Fig. 1. GTV (magenta contour), PTVsetup (red contour), and PTV (green contour) for patient 2. (Color version of figure is available online.)
Results The values of several dosimetric indices that are regularly used in our clinic to compare treatment plans and evaluate DVHs are shown for both 3D and 4D calculations in Tables 3 and 4. The results of a paired Student t-test comparing the values of these dosimetric indices corresponding to 3D and 4D DVHs are also shown in those tables. The dosimetric indices corresponding to the PTV show statistically significant differences when 3D and 4D DVHs are compared. For all patients, PTV dosimetric indices were lower in the 4D calculations than in the 3D calculations, as expected. Conversely, concerning the GTV and the PTVsetup, differences between the results of the 4D and the 3D calculations were minor, except in patient 7. The 4D calculations for patient 7 rendered an undercoverage of the PTVsetup when compared with the 3D calculations, which rose to a level that would be concerning for clinical effect (D994D ¼ 46.9 Gy and D993D ¼ 52.1 Gy, a difference of
5.2 Gy). The shape of this tumor was extremely irregular, and as such, the resulting ITV and PTV were also very irregular, as can be seen in Fig. 2. Excluding patient 7, the largest deficit of the D994D with respect to D993D in either the GTV or the PTVsetup was 1.8 Gy. Finally, using 3D and 4D DVHs corresponding to the GTV for each of the patients, the TCPs for 4 different dose schedules, 2 different sets of alpha and beta parameters, and 2 different tumor cell densities were calculated. As can be seen from Tables 5 and 6, the TCPs for the schedules representative of SBRT (18 Gy 3, 12.5 Gy 4, and 15 Gy 3) are almost always near enough to one for all different parameters such that at these dose levels and the TCP dose response curve are in the saturation zone for all patients, including patient 7. In this region, changes in the dose result in small changes in the TCP, as can be seen when the total dose is changed from 54 Gy to 45 Gy in Tables 5 and 6. For the schedule representative of a conventionally fractionated regimen, the TCP is nearly 0 for both tumor cell densities and the first set of parameters (α ¼ 0.19 Gy1 ⫾ 0.02 Gy1; β ¼ 0.02 Gy2 ⫾ 0.002 Gy2).
Table 2 Definitions of dosimetric indices used to approve and compare treatment plans Quantity
Region of interest
Description
Vp Vp Vp D95 D99 (GTV) D99 (PTVsetup) MD (GTV) MD (PTVsetup) Dmin Dmin Dmax Dmax
GTV PTVsetup PTV PTV GTV PTVsetup GTV PTVsetup GTV PTVsetup GTV PTVsetup
Fractional volume of GTV receiving at least the prescribed dose Fractional volume of PTVsetup receiving at least the prescribed dose Fractional volume of PTV receiving at least the prescribed dose Dose received by at least 95% of the PTV Dose received by at least 99% of the GTV Dose received by at least 99% of the PTVsetup Mean dose received by the GTV Mean dose received by the PTVsetup Minimum dose received by the GTV Minimum dose received by the PTVsetup Maximum dose received by the GTV Maximum dose received by the PTVsetup
MD ¼ mean dose.
G. Valdes et al. / Medical Dosimetry 40 (2015) 64–69
67
Table 3 Dosimetric indices used for plan approval and comparison Vp (GTV)
Patient
Vp (PTVsetup)
Vp (PTV)
D95 (PTV)
D99 (GTV)
D99 (PTVsetup)
3D
4D
3D
4D
3D
3D10
4D
3D
3D10
4D
3D
4D
3D
4D
1 2 3 4 5 6 7 8
98.3 100.0 100.0 100.0 100.0 100.0 98.4 100.0
97.9 100.0 99.9 100.0 100.0 100.0 78.6 100.0
92.5 97.9 96.0 94.5 100.0 97.7 89.2 91.0
89.8 99.4 96.2 96.9 95.7 98.6 56.9 97.6
95.3 96.0 96.7 94.0 95.1 93.6 89.1 91.8
95.7 95.7 96.1 95.6 79.2 84.1 81.3 97.3
70.2 58.7 78.4 72.6 81.0 74.8 38.6 75.1
54.1 54.3 55.0 53.3 50.1 53.4 53.7 52.8
54.4 54.4 54.5 54.4 48.4 50.9 52.5 55.6
46.1 26.3 43.7 37.7 48.4 38.2 40.8 38.9
52.8 55.6 59.0 56.8 53.1 56.1 53.9 56.6
52.3 56.8 59.4 61.6 51.7 56.5 52.5 58.7
48.3 53.3 43.3 49.5 50.9 53 52.1 50.1
47.2 54.4 44.7 51.2 49.1 53.7 46.9 52.5
Mean Paired t-test
99.6
97.0
94.8
91.4
93.9
90.6 S
68.7
53.3
53.1 S
40.0
55.5
56.2
50.1
NS
NS
50.0
NS
NS
3D10 is the 3D dose calculated using phase 10. NS ¼ not significant; S ¼ significant. Table 4 Dosimetric indices used for plan approval and comparison MD (GTV)
MD (PTVsetup)
Dmin (GTV)
Dmax (GTV)
Dmin (PTVsetup)
Dmax (PTVsetup)
Dmin (PTV)
3D
4D
3D
4D
3D
4D
3D
4D
3D
4D
3D
4D
3D
3D10
4D
3D
3D10
4D
60.8 57.4 62.9 66.7 55.2 60.3 55.8 60.2
58.7 58.3 62.7 67.4 54.0 59.1 54.6 62.3
59.4 56.7 60.9 64.0 53.9 58.7 55.3 58.2
57.5 57.4 60.8 62.7 52.6 57.8 54.0 60.5
49.8 55.3 53.3 54.1 52.3 56.1 53.4 53.6
48.7 56.3 53.5 59.0 50.9 55.6 51.9 55.1
63.7 59.2 65.6 72.1 57.2 62.8 57.7 62.9
60.9 59.7 65.2 69.9 55.6 61.5 56.4 64.9
39.3 50.7 34.7 39.6 49.6 48.9 48.4 45.8
37.7 49.7 34.5 44.1 47.5 52.1 43.6 45.3
64.1 59.3 65.6 72.1 57.3 62.8 59.2 63.9
61.0 59.8 65.2 69.9 55.6 61.4 57.5 64.9
41.4 45.9 36.8 41.3 45.5 42.3 49.2 43.0
43.1 44.8 36.6 41.1 44.1 40.5 47.9 45.6
31.9 14.3 28.8 23.2 44.4 24.5 25.3 19.5
64.1 60.4 65.6 75.4 57.3 63.8 60.2 65.8
65.3 68.0 66.9 79.8 56.3 62.9 58.8 69.2
60.9 59.7 65.2 69.9 55.6 61.5 57.5 65.0
59.6
58.4
57.9
53.5
53.9
62.7
61.8
44.6
44.3
63.0
61.9
43.2
43.0 S
26.5
64.1
65.9 S
61.9
59.9 NS
NS
NS
NS
NS
S
Dmax (PTV)
MD ¼ mean dose; NS ¼ not significant; S ¼ significant.
Fig. 2. Irregularly shaped PTV of patient 7. The PTV is shown by the red contour and 99% isodose line according to the 4D calculation is shown by the green contour. (Color version of figure is available online.)
68
G. Valdes et al. / Medical Dosimetry 40 (2015) 64–69
Table 5 TCP calculated for α ¼ 0.19 ⫾ 0.02, β ¼ 0.02 ⫾ 0.002, 4 different dose schedules and 2 different cell densities Patients
Tumor cell density (cells/cm3)
TCP_54 Gy_3 3D
7
1 2 3 4 5* 6 7 8
Mean
n
4D
10 108 107 108 107 108 107 108 107 108 107 108 107 108 107 108
4 4 4 4 4 4 4 4 4
107 108
4 0.99 4 0.99
4 4 4 4 4 4
TCP_45 Gy_3
0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.96 0.99 0.99 0.99 0.99 0.99 0.99
4 4 4 4 4 4 4
4 4 4 4 4 4
3D 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.91 0.99 0.99 0.99 0.99 0.99 0.99
4 0.99 4 0.99
4 4 4 4 4
4 4 4 4
TCP_66 Gy_33 4D
0.99 0.91 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.95 0.99 0.99 0.99 0.98 0.99 0.95
4 0.99 0.97
4 4 4 4 4 4
4 4 4 4
3D 0.98 0.86 0.99 0.99 0.99 0.99 0.99 0.99 0.98 0.91 0.99 0.99 0.99 0.97 0.99 0.98
4 0.99 0.96
o o o o o o o o o o o o o o o o
4D 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01
o 0.01 o 0.01
o o o o o o o o o o o o o o o o
0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01
o 0.01 o 0.01
Patient 5 was prescribed 50 Gy in 4 fractions instead of 54 Gy in 3 fractions as it is customary in our clinic for central lung tumors.
The TCP has mean values of 0.73 and 0.70 for 3D and 4D DVHs, respectively, when the second set of parameters (α ¼ 0.30 Gy1 ⫾ 0.02 Gy1; β ¼ 0.02 Gy2 ⫾ 0.002 Gy2) is used. However, no statistically significant differences were found between TCPs calculated using 3D or 4D DVHs for SBRT or regular fractionated regimes. It should be noted that, for some tumors, the TCP is larger under the 4D calculations. That may be because of the effect of respiratory motion smoothing over relatively cold areas in these tumors.
the PTV. However, the results of these calculations on the PTV are an indication of the dose received by the tissue under the PTV in the reference CT image where it was contoured, rather than the dose received by the physical area where the tumor moves and which is targeted in the plan. Therefore, when 4D calculations are performed, uniform coverage of the accumulated 4D dose on the tissue region defined by the PTV in the reference CT image should not be sought, as it would result in an unnecessary dose to the patient. Conversely, evaluating 4D dose using only the GTV is not correct either because the GTV does not contain a margin for setup variations that are expected to occur according to a statistical probability distribution. For that reason, we introduced the concept of the PTVsetup for evaluation of 4D accumulated dose. The PTVsetup is an expansion of the GTV that includes margin for setup uncertainty but not for tumor motion. We chose a margin of 4 mm, based on our institution's experience with cone-beam CT image guidance for soft tissue targets in the context of SBRT.
Discussion We found significant differences in D95(PTV) when 3D dose calculations were compared with 4D dose calculations for 8 Stage I-II lung cancer radiotherapy plans. These results are similar to those found by Starkschall et al.3; however, our interpretation of them is different. The 4D calculations indicate an undercoverage of
Table 6 TCP calculated for α ¼ 0.30 ⫾ 0.02, β ¼ 0.02 ⫾ 0.002, 4 different dose schedules and 2 different cell densities Patients
Tumor cell density (cells/cm3)
TCP_54 Gy_3 3D
1 2 3 4 5* 6 7 8 Mean
n
TCP_45 Gy_3 4D
107 108 107 108 107 108 107 108 107 108 107 108 107 108 107 108
o o o o o o o o o o o o o o o o
107 108
o 0.99 o 0.99
0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99
o o o o o o o o o o o o o o o o
3D 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99
o 0.99 o 0.99
o o o o o o o o o o o o o o o o
TCP_66 Gy_33 4D
0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99
o 0.99 o 0.99
3D
4D
0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99
0.66 0.19 0.72 0.25 0.96 0.77 0.97 0.80 0.69 0.24 0.82 0.40 0.68 0.21 0.74 0.30
0.53 0.11 0.78 0.34 0.96 0.76 0.99 0.92 0.56 0.15 0.75 0.30 0.58 0.13 0.86 0.48
o 0.99 o 0.99
0.73 0.37
0.70 0.38
o o o o o o o o o o o o o o o o
Patient 5 was prescribed 50 Gy in 4 fractions instead of 54 Gy in 3 fractions as it is customary in our clinic for central lung tumors.
G. Valdes et al. / Medical Dosimetry 40 (2015) 64–69
Despite the dosimetric differences related to the PTV, when the 3D and 4D dose calculations corresponding to the GTV or the PTVsetup were analyzed, only 1 patient showed a dosimetric difference that rose to a level of clinical significance. We regarded a dose deficit of 5% or greater of the prescription dose as one that would be regarded as clinically significant in most clinics. In the case of patient 7, a 9% deficit in D99(PTVsetup) in the 4D vs the 3D calculations occurred. We hypothesize that such differences were seen because of to the irregular shape of the tumor and PTV (Fig. 2). No other dosimetric deficit was larger than 4% of the prescription dose. On the contrary, when TCP was calculated for these tumors using a wide range of alpha, beta, and cell density parameters taken from the literature, TCP for SBRT fractionations used in our clinic (18 Gy 3 for peripheral tumors and 12.5 Gy 4 for central tumors), TCP was almost always greater than 99% and always greater than 90%. For these SBRT fractionations, large changes in the total dose received by the GTV have negligible effect on the TCP because the TCP response curve is in the saturation zone. These analytical results are consistent with the published reports of lung SBRT that show that, regardless of tumor location or tumor motion, when adequate dose is applied and good image guidance is employed, local control has been in excess of 90%.17,18 To add more complexity to the discussion, it is important to highlight that interplay effects between the IMRT leaf movement and the respiration cycle were not included in our 4D dose calculation method. According to Rao et al.,1 calculating the 4D dose as described in this study offers similar results as taking into account the interplay between the IMRT leaf movement and the respiration cycle. This statement is equivalent to that made by Bortfeld et al.19 who showed that the reference dose for the case of intrafraction motion should be the average dose obtained over the path of the motion. However, as has been shown by Seco et al., these statements are only valid in general for those plans where the delivery time of each of the IMRT segments is an order of magnitude bigger than the breathing period (3 to 5 seconds). If the delivery time of each of the IMRT segments is on the same order of magnitude of the breathing period, the interplay effect could cause nonnegligible effects.20 It is important to highlight that the results presented here were obtained for a sample of patients with small and highly mobile tumors. The mean tumor excursion of our 8 patients was 21 mm, with a standard deviation of 11 mm, while the mean and RMS tumor volumes were 12 ⫾ 8.8 cm3. In comparison, Rao et al.1 studied 10 patients with a mean tumor excursion of 14 ⫾ 6.6 mm and a mean volume of 42 ⫾ 49 cm3, whereas Starkschall et al.3 studied 15 patients with a mean excursion of 7.0 ⫾ 6.0 mm and a mean volume of 107 ⫾ 109 cm3. We expect that on average, small tumors will show a larger difference between 3D and 4D calculations, although individual cases may vary substantially in their sensitivities to 4D calculations. A small tumor with large motion corresponds to a PTV whose volume is a large multiple of the GTV, at least relatively, when compared with a large tumor with small motion. Conventionally, the plan is normalized so that the prescription dose covers 95% of the PTV. The portion of the PTV not receiving the prescription dose is normally in the boundary region of the PTV. This boundary region of the PTV is proportionally larger compared with the GTV in a small tumor that moves to a large degree, compared to a large and relatively stationary tumor.
Conclusions Summarizing, it is likely that for lung stereotactic body radiotherapy with doses of 18 Gy 3 fractions prescribed to 95% of the
69
PTV, 3D calculations are adequate for treatment optimization as the DVHs estimated using this method produce similar TCP as those obtained using 4D calculations. However, if the resources are available, 4D calculations should be carried out especially for highly mobile tumors with irregular shapes. Furthermore, if 4D calculations are performed, the dose received by the PTVsetup, i.e., the GTV with a margin to account for setup uncertainty but not respiratory motion, should be evaluated in addition to evaluating dose to the GTV and the PTV.
Supplementary Materials Supplementary material cited in this article is available online at http://dx.doi.org/10.1016/j.meddos.2014.10.002.
References 1. Rao, M.; Wu, J.; Cao, D.; et al. Dosimetric impact of breathing motion in lung stereotactic body radiotherapy treatment using image-modulated radiotherapy and volumetric modulated arc therapy. Int. J. Radiat. Oncol. Biol. Phys. 83: e251–e256; 2012. 2. Rietzel, E.; Chen, G.T.; Choi, N.C.; et al. Four-dimensional image-based treatment planning: target volume segmentation and dose calculation in the presence of respiratory motion. Int. J. Radiat. Oncol. Biol. Phys. 61:1535–50; 2005. 3. Starkschall, G.; Britton, K.; McAleer, M.F.; et al. Potential dosimetric benefits of four-dimensional radiation treatment planning. Int. J. Radiat. Oncol. Biol. Phys. 73:1560–5; 2009. 4. Nagata, Y.; Wulf, J.; Lax, I.; et al. Stereotactic radiotherapy of primary lung cancer and other targets: results of consultant meeting of the International Atomic Energy Agency. Int. J. Radiat. Oncol. Biol. Phys. 79:660–9; 2011. 5. Low, D.A.; Nystrom, M.; Kalinin, E.; et al. A method for the reconstruction of four-dimensional synchronized CT scans acquired during free breathing. Med. Phys. 30:1254–63; 2003. 6. Hodapp, N. The ICRU Report 83: prescribing, recording and reporting photonbeam intensity-modulated radiation therapy (IMRT)]. Strahlenther. Onkol. 188:97–9; 2012. 7. Flampouri, S.; Jiang, S.B.; Sharp, G.C.; et al. Estimation of the delivered patient dose in lung IMRT treatment based on deformable registration of 4D-CT data and Monte Carlo simulations. Phys. Med. Biol. 51:2763–79; 2006. 8. Britton, K.R.; Starkschall, G.; Liu, H.; et al. Consequences of anatomic changes and respiratory motion on radiation dose distributions in conformal radiotherapy for locally advanced non–small-cell lung cancer. Int. J. Radiat. Oncol. Biol. Phys. 73:94–102; 2009. 9. Brock, K.K.; McShan, D.L.; Ten Haken, R.K.; et al. Inclusion of organ deformation in dose calculations. Med. Phys. 30:290–5; 2003. 10. Wang, H.; Dong, L.; O’Daniel, J.; et al. Validation of an accelerated ‘demons’ algorithm for deformable image registration in radiation therapy. Phys. Med. Biol. 50:2887–905; 2005. 11. Chi, A.; Liao, Z.; Nguyen, N.P.; et al. Systemic review of the patterns of failure following stereotactic body radiation therapy in early-stage non–small-cell lung cancer: clinical implications. Radiother. Oncol. 94:1–11; 2010. 12. El Sharouni, S.Y.; Kal, H.B.; Battermann, J.J. Tumour control probability of stage III inoperable non–small cell lung tumours after sequential chemoradiotherapy. Anticancer Res. 25:4655–61; 2005. 13. Niemierko, A.; Goitein, M. Implementation of a model for estimating tumor control probability for an inhomogeneously irradiated tumor. Radiother. Oncol. 29:140–7; 1993. 14. Tome, W.A.; Fowler, J.F. On cold spots in tumor subvolumes. Med. Phys. 29:1590–8; 2002. 15. Webb, S.; Nahum, A.E. A model for calculating tumour control probability in radiotherapy including the effects of inhomogeneous distributions of dose and clonogenic cell density. Phys. Med. Biol. 38:653–66; 1993. 16. Qi, X.; Low, D.A.; Kupelian, P.; et al. Analysis of outcomes in early stage non small cell lung cancer irradiation: fractionation scheme implications for stage I versus stage II disease. Int. J. Radiat. Oncol. Biol. Phys. 84(3):S566; 2012. 17. Timmerman, R.; Bastasch, M.; Saha, D.; et al. Stereotactic body radiation therapy: normal tissue and tumor control effects with large dose per fraction. Front. Radiat. Ther. Oncol. 43:382–94; 2011. 18. van Baardwijk, A.; Tomé, W.A.; van Elmpt, W.; et al. Is high-dose stereotactic body radiotherapy (SBRT) for stage I non–small cell lung cancer (NSCLC) overkill? A systematic review Radiother. Oncol. 105:145–9; 2012. 19. Bortfeld, T.; Jokivarsi, K.; Goitein, M.; et al. Effects of intra-fraction motion on IMRT dose delivery: statistical analysis and simulation. Phys. Med. Biol. 47:2203–20; 2002. 20. Seco, J.; Sharp, G.C.; Turcotte, J.; et al. Effects of organ motion on IMRT treatments with segments of few monitor units. Med. Phys. 34:923–34; 2007.