Evaluation of the breath-hold approach in proton therapy of lung tumors

Proton therapy has the potential to improve the local control rates and reduce the risk of toxicity for
lung cancer patients. However, the delivery of proton therapy is prone to uncertainties caused by
anatomical changes and motion during the treatment and between the treatment fractions which
may compromise its effectiveness. The dosimetric uncertainty of lung cancer proton therapy can be
minimized through the use of motion mitigation techniques; increased margins, beam gating and
breath-hold, tumor tracking and rescanning. Most of these techniques have been extensively
investigated in the literature showing good results, but the breath-hold technique has remained
relatively unexplored. The breath-hold technique has shown promising stability and reproducibility,
together with good patient compliance, from photon radiation therapy treatment. The aim of this
thesis was to investigate the robustness of the breath-hold approach for pencil beam scanned (PBS)
proton therapy. More specifically, the residual motion as seen on repeated breath-hold computed
tomography (CT) scans and fluoroscopy acquisitions were investigated using simulation and
experimental studies. The work was carried out at Paul Scherrer Institute (PSI), Switzerland with
clinical data from the Rigshospitalet, Copenhagen University Hospital, Denmark. All treatment plans
were constructed in the in-house treatment planning system (TPS) PSIplan at PSI. Four dimensional
(4D) dose simulations have been carried out using our in-house developed 4D dose calculation
software based on the PSIplan and Plastimatch image registration. Experimental studies were
performed at PSI using a dynamic anthropomorphic breathing phantom whose artificial lungs can be
deflated and inflated according to pre-programmed motion patterns e.g. from patients. The work
included in this thesis was divided into four different studies and the summarized results are as
follows:
• Study 1: The robustness to inter-fractional breath-hold motion is sufficient for 14/15 single
field uniform dose (SFUD) cases for patients diagnosed with early-stage non-small cell lung
cancer (NSCLC). Patients with small tumors and large baseline shift are prone to achieve
larger dose deviations.
• Study 2: The robustness to intra- and inter-fractional breath-hold motion is sufficient for
12/15 intensity modulated proton therapy (IMPT) cases of patients diagnosed with locallyadvanced
NSCLC. The change in water-equivalent path length (WEPL) is shown to be a good
predictor of plan robustness, in addition to baseline shifts of the tumor.
• Study 3: Robust beam angles are patient dependent and the change in WEPL is confirmed to
be correlated to dose degradation.
• Study 4: The robustness to intra-breath-hold motion is sufficient for single field uniform dose
(SFUD) plans, but not for IMPT.
Further studies are necessary to identify more parameters important for prediction of plan
robustness, through e.g. more extensive image data sets. Incorporating this motion information in
the 4D dose calculation software would strengthen the simulation method. Improving the dosimeter
for the LuCa phantom could increase the precision of the dose readout for the experimental studies.
Overall, the results of this thesis encourage a clinical implementation of the breath-hold approach.
The clinical benefit of this approach remains to be discovered.