Minimizing normal tissue dose spillage via broad-range optimization of hundreds of intensity modulated beams for treating multiple brain targets

Abstract

Variable normal tissue dose and inter-target dose interplay effects have been reported in volumetric modulated arc therapy (VMAT) of multiple brain metastases. In order to minimize such adverse effects, a Broad-Range Optimization of Modulated Beam Approach (BROOMBA) was developed whereby hundreds of intensity-modulated beams surrounding the central axis of the skull were progressively selected and optimized. To investigate technical feasibility and potential dosimetric benefits of BROOMBA, we first developed such an approach on a standalone workstation and then implemented it for a multi-center benchmark case involving 3 to 12 multiple brain metastases. The BROOMBA planning results was compared with VMAT treatment plans of the same case using coplanar and non-coplanar arc beams. We have found that BROOMBA consistently outperformed VMAT plans in terms of low-level normal brain sparing and reduction in the dose interplay effects among the targets. For example, when planning simultaneous treatment of 12 targets, BROOMBA lowered the normal brain dose by as much as 65% versus conventional VMAT treatment plans and the dose interplay effects across 8 Gy to 12 Gy levels was reduced to be negligible. In conclusion, we have demonstrated BROOMBA as a powerful tool for improving the planning quality of multiple brain metastases treatments via modern high-output linear accelerators.

Keywords: beam optimization; brain metastases; dose interplay; intensity-modulated arc therapy.

Figure 1

Illustration of beam/arc orientations and pathways in terms of gantry and couch angels for a treatment planning case (n=3 targets) between the noncoplanar VMAT results and the noncoplanar BROOMBA results (1) shows the BROOMBA beam configurations (N=20), (2) shows a 4-arc VMAT beam arrangement.

Figure 2

Comparison of normal brain at different peripheral isodose levels between coplanar and non-coplanar BROOMBA and VMAT techniques for increasing target numbers (e.g., N=3, 6, 9, and 12, respectively).

Figure 3

Comparison of low-level normal brain spillage isodose distributions for the case of N=3 (panel a) and N=12 (panel b) for the 6 MV x-ray treatment plans. In both panel (a) and (b), the top figures are the noncoplanar 4-arc VMAT treatment plans and bottom figures are the BROOMBA treatment plans via 20 non-coplanar beams. Note the area difference in the low dose levels between the two techniques.

Figure 4

Dependence of dose interplay effects (i.e. mean isodose volume per target with increasing number of targets) at the 8-Gy and the 12-Gy isodose levels. Note the differences in the y-intercepts and slopes among these curves and the lowest values produced by the non-coplanar BROOMBA technique.

Figure 5

Example of the peripheral 8-Gy isodose volume dependence on the increasing number of beams as determined by the BROOMBA for treating different number of targets of N=3, 6, 9, and 12. The curve continuously decreased with increasing number of beams for BROOMBA and the maximum gain was approximately reached near 20-30 beams depending on the number of targets under consideration.