Session Item

For proton treatment planning, one of the main causes of range uncertainty is on the CT-based estimation of stopping power ratio (SPR) relative to water. SPR estimation is usually based on single energy CT (SECT) scans, however, previous studies have shown that dual energy CT (DECT) can result in a more accurate SPR estimation and thereby a reduction of range uncertainty. In robust optimization, typically setup and range uncertainties are accounted for, and lead to an expansion of the irradiated volume to assure target coverage in all error scenarios. This study investigated how large a range uncertainty reduction is needed to obtain a clinically relevant dose sparing to organs at risk (OARs) for patients with brain tumours. 

This study included treatment planning CT scan of ten patients with brain tumours acquired in Twin Beam mode at a SOMATOM Definition Edge CT scanner (Siemens Healthineers, Forcheim, Germany). The patients were treated with spot scanned proton therapy and a prescription dose of 50.4Gy in 28 fractions. Based on the field directions and optimization objectives of the original treatment plan, new plans were robustly optimized with reduced range uncertainties using Eclipse treatment planning system (Varian, Palo Alto, CA). Seven plans were created for each patient, with a range uncertainty of 3.5% (original plan), 3.0%, 2.5%, 2.0%, 1.5%, 1.0% and 0.0%, respectively. For all robust optimizations, a setup error of 2mm was used. Each plan was optimized until a clinically acceptable plan (D95%>98% for clinical target volume (CTV)) was obtained for all fourteen setup and range scenarios. The plans were normalized to CTV mean dose. Dosimetric effect of a reduced range uncertainty was evaluated for OARs and ring structures of various thickness (from 1mm to 20mm) surrounding the CTV. Comparisons were made by evaluating mean and max dose to OARs and ring structures. 

The results show that reducing the range uncertainty in the treatment plan gives a slight reduction in nominal dose to the surrounding tissue. Mean dose to brain outside CTV (Brain-CTV) is on average reduced by 3.4%(std=0.2%) over the ten patients when reducing range uncertainty from 3.5% to 2%. Dose to the ring structure of 10mm surrounding the CTV was reduced by 2% on average. In the worst-case scenarios for Brainstem and Chiasm, reducing the range from 3.5% to 2.0% results in a reduction in max dose by 4.1% (0.7Gy) and 4.7% (1.5Gy) respectively. Table 1 contains the average dose differences for the evaluated OARs.

Reducing the range uncertainty leads to meaningful reductions in dose to OARs. Whether the reduction is clinically relevant or not needs to be seen in a context; if the dose to OARs is close to the dose constraints maximum, the reductions shown in this study are very relevant. To obtain a relevant dose reduction of 2% in the mean dose to the brain outside CTV, new modalities such as DECT needs to ensure that the range uncertainty can be reduced down to 2-2.5%.