Reduction of the secondary neutron dose in passively scattered proton radiotherapy, using an optimized pre-collimator/collimator

Abstract

Proton radiotherapy represents a potential major advance in cancer therapy. Most current proton beams are spread out to cover the tumor using passive scattering and collimation, resulting in an extra whole-body high-energy neutron dose, primarily from proton interactions with the final collimator. There is considerable uncertainty as to the carcinogenic potential of low doses of high-energy neutrons, and thus we investigate whether this neutron dose can be significantly reduced without major modifications to passively scattered proton beam lines. Our goal is to optimize the design features of a patient-specific collimator or pre-collimator/collimator assembly. There are a number of often contradictory design features, in terms of geometry and material, involved in an optimal design. For example, plastic or hybrid plastic/metal collimators have a number of advantages. We quantify these design issues, and investigate the practical balances that can be achieved to significantly reduce the neutron dose without major alterations to the beamline design or function. Given that the majority of proton therapy treatments, at least for the next few years, will use passive scattering techniques, reducing the associated neutron-related risks by simple modifications of the collimator assembly design is a desirable goal.

Figure 1

Geometry for the simplified Monte Carlo simulations. A uniform mono-energetic 235 MeV proton beam is incident on the face of the pre-collimator, which is upstream of the patient-specific collimator which is, in turn, immediately upstream of a cylindrical homogeneous tissue-equivalent phantom (to the right, not shown). The pre-collimator prevents unnecessary protons from impinging on the patient-specific collimator, while the final collimator provides the patient-specific collimation, but also shields the patient from neutrons produced in the pre-collimator. For these calculations, both the collimator and pre-collimator had circular internal and external diameters, respectively, of 50 and 113 mm. Doses due to neutrons originating in the collimator were tallied in 10 mm thick slices along the length of a cylindrical homogeneous tissue-equivalent phantom (1.5 m length × 0.5 m diameter), axis perpendicular to the proton-beam direction, located immediately downstream of the patient-specific collimator.

Figure 2

Simulations of the proton therapy treatment heads at the gantry beamlines at MGH, with a 65-mm thick brass collimator (above), and an optimized plastic or metal/plastic collimator (below). The proton beam is incident from the left, and the modulated collimated beam is incident on a water phantom on the right. Beam and treatment-head settings were chosen to represent a typical field, with a prescribed range of 157 mm, a modulation width of 68.5 mm, and an aperture opening diameter of 60 mm. The area irradiated with protons at the base plate position corresponds to a diameter of roughly 250 mm, depending on the setting of variable collimators upstream of the snout retraction area. The treatment head components are described in detail elsewhere (Paganetti et al 2004).

Figure 3

Calculated patient dose (per proton incident on the face of the collimator) from neutrons originating in patient-specific collimators made out of different materials. Geometry was as in figure 1, with no pre-collimator. The doses were tallied in a cylindrical tissue-equivalent phantom, axis perpendicular to the beam axis, located immediately downstream of the collimator, as a function of distance from the proton beam axis. Typical statistical uncertainties (1σ) in the results related to the Monte Carlo simulation are ±0.3% on axis, and ±1% at 0.7 m off axis. The thicknesses of each of the collimators correspond to the range of 235 MeV protons in that collimator material (see table 1). Also shown is the `direct' neutron dose arising from proton interactions in the tissue phantom itself.

Figure 4

Calculated peak patient dose (per proton incident on the face of the collimator) from neutrons originating in a brass collimator of various thicknesses. Geometry as in figures 1 and 3 with no pre-collimator. Note the minimum brass collimator thickness (i.e., to stop 235 MeV protons) is 65 mm.

Figure 5

Calculated lateral penumbra widths (beam width at 80% of maximum dose minus beam width at 20% of maximum dose) for a 65 mm thick brass and a 200 mm SWX-207HD polyethylene collimator (aperture opening diameter 60 mm), using the detailed beamline geometry shown in figure 2, and typical beam and nozzle settings. In the entrance region, the lateral penumbra is worse from the thicker plastic collimator, but in the spread-out Bragg peak (in this case, from 110 to 150 mm), the difference in the lateral penumbra is 0.5 mm.

Figure 6

Calculated patient dose (per proton incident on the face of the collimator) from neutrons originating in a pre-collimator + collimator configuration (see figure 1). In each case the patient-specific collimator was 65-mm thick brass; the four pre-collimators were each of sufficient thickness to stop 235 MeV protons (see table 1), and were located immediately upstream of the patient-specific collimator.

Figure 7

Calculated peak patient dose (per proton incident on the face of the collimator) from neutrons originating in a pre-collimator + collimator configuration (see figure 1), as a function of the thickness of the pre-collimator. In each case the patient-specific collimator was 65 mm thick brass, and the pre-collimator was located immediately upstream of the patient-specific collimator. The points correspond to the pre-collimator thickness sufficient to stop 235 MeV protons (as in figure 6).

Figure 8

Calculated patient dose (per proton incident on the face of the collimator) from neutrons originating in a pre-collimator + brass collimator configuration (see figure 1), for different pre-collimator/collimator separations. In each case the collimator was 65 mm thick brass. The pre-collimator thicknesses were as in table 1 and figure 6.