ESTRO 2022 - Abstract Book

S1539

Abstract book

ESTRO 2022

20 left breast cancer patients, in DIBH, were planned for VMAT technique with the Varian Eclipse TPS to deliver 40.5 Gy in 15 fractions to the whole breast. Two planning strategies were adopted: PartArc , with 2 partial arcs about 200° long; and AvoidArc , with the same partial arcs and additional avoidance sectors (arc sectors with 0 dose rate) set in the PO to reduce the dose to contralateral breast and lung. RapidPlan, RP, models were generated based on the two plan sets, named RP_Part and RP_Avoid . A new set of plans was optimized using only the two RP models. The beam geometry was the same as the initial plans, without setting avoidance sectors inside the optimizer. The RP generated plans were compared with the original initial plans. In a second part of the study a third RP model was generated, with plans similar to the AvoidArc , but forcing the beam geometry to the avoidance sectors ( 2+2Short plans): the two partial arcs were split in two short arcs each. The model was named RP_2+2. The plan differences between the AvoidArc and the 2+2Short plans were minimal and not significant since they were in practice using the same geometry, the first set inside the optimizer as dose rate setting on a sector, the second set as beam geometry with split shorter arcs. Plans generated with RP_2+2 , with the geometry without splitting the partial arcs, were compared with the AvoidArc plans to determine if the initial beam geometry information in the RP model can improve the PO performances. Results The target coverage and dose homogeneity were acceptable in all cases, with manual and RP based plans. The following table reports the mean doses, averaged over the whole patient’s cohort, and standard deviation to the homolateral structures, left lung and heart, and the contralateral structures, right breast and lung, for the manual plans PartArc and AvoidArc , and for the RP generated RP_Part , RP_Avoid and RP_2+2 .

The PO, using the RP_Avoid , was unable to reduce the dose rate in the avoidance sectors when they were not manually defined, although the RP_Avoid estimated DVHs of the contralateral structures were in agreement with the AvoidArc plans. Similarly, in the case of the RP model generated with the split short arcs, the PO did not reduce the contralateral doses. Only in the case with the arcs split into short arcs in the initial geometry, the dose to the contralateral structures was correctly achieved. The last row in the table reports the case of RP_2+2 and split arcs as initial geometry, where the contralateral structures are similar to the AvoidArc and 2+2Short cases. Conclusion The PO showed to be unable to modulate the dose rate enough to produce the RP DVH estimation. In particular, it does not reduce the dose rate in the beam toward structures where a very low dose (scattering) is required.

PO-1737 Robustness of the pseudo-skin flash in VMAT breast radiotherapy with Acuros XB algorithm

A. Honkanen 1 , L. Porra 1 , A. Rintala 1 , V. Reijonen 1 , T. Seppälä 1 , M. Tenhunen 1

1 Helsinki University Hospital, Comprehensive Cancer Center, Helsinki, Finland

Purpose or Objective Deformation of the surface of the breast due to e.g. , swelling and patient positioning is often observed during the external radiotherapy. The robustness of the treatment plan can be increased by expanding the field apertures to include an air margin. One approach to perform this in VMAT planning is to expand the original PTV outside the skin and fill it with an expanded body structure or virtual bolus; this method is referred to as pseudo skin flash. The optimal features of the bolus and expanded PTV has been investigated earlier [1,2]. However, dose calculations in these studies were done with the AAA algorithm which is known to underperform in the regions with large inhomogeneities in density such as the skin-air interface, compared to the Acuros XB (AXB) algorithm [3]. The purpose of our study was to investigate the optimization of the bolus thickness and HU value selection to produce pseudo skin flash in VMAT breast plans using AXB dose calculation algorithm. Materials and Methods A test patient with a left breast PTV (no lymph nodes) was used to study the effect of pseudo skin flash: two back-and- forth VMAT field pairs for 6 MV photons were set at 282°-355° and 95°-175° with collimator rotations at 10° and 350°. The virtual boli used in the study were 5, 10, and 15 mm in thickness with the CT numbers of -1000, -500, 0, 100, and 250 HU. The VMAT plans were optimized to an extended PTV (PTV + bolus - 5 mm margin) with criteria for the extended PTV and NTO only, and doses were calculated using AXB in Eclipse treatment planning system (v.16.1, Varian Medical Systems, CA, USA). The robustness of the plans was examined by a) calculating the doses with ±7 mm isocenter shifts along the main axes and b) using a 7 mm thick bolus of -100 HU to simulate swelling.

Results