Four-dimensional Plan Optimization for the Treatment of Lung Tumors Using Pencil-beam Scanning Proton Radiotherapy

Abstract

Purpose This study aimed to evaluate the effectiveness of four-dimensional (4D) robust optimization for proton pencil-beam scanning (PBS) treatment of lung tumors. Patients and methods In seven patients with lung cancer, proton beam therapy was planned using 4D robust optimization over 4D computed tomography (CT) data sets. The gross target volume (GTV) was contoured based on individual breathing phases, and a 5-mm expansion was used to generate the clinical target volume (CTV) for each phase. The 4D optimization was conducted directly on the 4D CT data set. The robust optimization settings included a CT Hounsfield unit (HU) uncertainty of 4% and a setup uncertainty of 5 mm to obtain the CTV. Additional target dose objectives such as those for the internal target volume (ITV) as well as the organ-at-risk (OAR) dose requirements were placed on the average CT. For comparison, three-dimensional (3D) robust optimization was also performed on the average CT. An additional verification 4D CT was performed to verify plan robustness against inter-fractional variations. Results Target coverages were generally higher for 4D optimized plans. The difference was most pronounced for ITV V70Gy when evaluating individual breathing phases. The 4D optimized plans were shown to be able to maintain the ITV coverage at full prescription, while 3D optimized plans could not. More importantly, this difference in ITV V70Gy between the 4D and 3D optimized plans was also consistently observed when evaluating the verification 4D CT, indicating that the 4D optimized plans were more robust against inter-fractional variations. Less difference was seen between the 4D and 3D optimized plans in the lungs criteria: V5Gy and V20Gy. Conclusion The proton PBS treatment plans optimized directly on the 4D CT were shown to be more robust when compared to those optimized on a regular 3D CT. Robust 4D optimization can improve the target coverage for the proton PBS lung treatments.

Keywords: 4d robust optimization; interplay effect; proton pbs treatment; robust optimization.

Figure 1. Target contour definition

The ITV (green) and IGTV (red) contours as observed in the (a) maximal inspiration phase, (b) average CT scan and (c) maximal expiration phase of the breathing cycle. Note that the IGTV has been contoured directly on the MIP and transferred to the individual breathing phase CTs. ITV: internal target volume, IGTV: internal gross target volume, MIP: maximum intensity projection

Figure 2. Dose-volume histogram comparison

Dosimetric comparison between 3D and 4D robust optimized plans. All plans are normalized to ITV D99% at 70 Gy(RBE) on the average CT, as seen in (h) and (k). The 4D optimized plans generally show better target coverage with negligible lung difference. 3D: three-dimensional, 4D: four-dimensional, ITV: internal target volume, CT: computed tomography

Figure 3. Dose distribution comparison

Comparison between 3D robust optimization (a, b, c) and 4D robust optimization (d, e, f) on the planning CT. The 3D plan is optimized on the average CT (b), and calculated on the maximal inspiration phase (a) and the maximal expiration phase (c). The 4D plan is optimized simultaneously on the maximal inspiration phase (d), the average CT (e), and the maximal expiration phase (f). Note that the 4D plan is optimized simultaneously on the average CT, the maximal inspiration phase, the maximal expiration phase, and an intermediate breathing phase (not shown). The contour shown in all panels is the ITV. The 4D optimized plan shows more consistent ITV coverage. Lung sparing between these two plans are comparable (details in text). 3D: three-dimensional, 4D: four-dimensional, CT: computed tomography, ITV: internal target volume

Figure 4. Verification computed tomography: dose-volume histogram

Evaluation of 3D vs. 4D plan robustness using VFCT scans. The internal target volume (ITV) V70Gy is used as the criterion. The 3D optimized plans are shown in the top row (a, b, c), and the 4D optimized plans in the bottom row (d, e, f). The 3D plans were robustly optimized on the average CT and evaluated on the maximum inspiration and maximum expiration phases of the planning CTs. The 3D plan’s target coverage is indeed ensured on the average planning CT (b), but is not guaranteed on the individual breathing phases of the planning CTs (a) and (c). Specifically, for patient 6, the 3D plan’s target coverage is less than 80% on the maximum inspiration and maximum expiration phases of the planning CT, while for patient 7, the 3D plan has >98% coverage. More importantly, the 3D plan’s target coverage, when evaluated on the VFCT, was consistently lower than 80% for both patient 6 and 7, even on the average VFCT. On the contrary, 4D optimized plans showed consistent target coverage on all phases of the planning CT as well as the VFCT (d, e, f). While good target coverage in all phases of the planning CT is indeed expected of the 4D optimized plans since it is specifically optimized on those specific CTs, these 4D optimized plans also demonstrated better target coverage across all phases in the VFCT. This suggested that 4D optimized plans are more robust against inter-fractional motions than 3D optimized plans. 3D: three-dimensional, 4D: four-dimensional, VFCT: verification computed tomography, CT: computer tomography

Figure 5. Verification computed tomography: dose distribution

Comparison between the 3D optimized plan (a, b, c) and 4D optimized plan (d, e, f), when calculated on the VFCT. The 3D optimized plan is calculated on the VFCT's maximal inspiration phase (a), average (b), and the maximal expiration phase (c). The 4D optimized plan is calculated on the same VFCT scans, i.e., maximal inspiration (d), average (e), and maximal expiration (f). Note that the 3D plan’s target coverage deteriorated on all the VFCT scans, while that of the 4D optimized plan only showed minor degradation. This shows that 4D optimized plans are indeed more robust against inter-fractional variations. This is most likely due to the fact that 4D optimized plans have more stringent robustness criteria when being optimized (details in text). 3D: three-dimensional, 4D; four-dimensional, VFCT: verification computed tomography