245 / 1082
S232
ESTRO 36 2017
_______________________________________________________________________________________________
Purpose or Objective
In the Netherlands, a subgroup of patients eligible for
proton therapy will be selected using the model-based
approach. So far, patients were selected by comparing
robustly optimized proton plans to non-robustly optimized
photon plans. However, using non-robust optimization for
photon plans may not explore the full potential of this
technique, possibly leading to an unfair comparison. The
main objective of this study was to investigate the benefit
of robust optimization for head and neck using photon
techniques.
Material and Methods
A cohort of nine head and neck cancer patients clinically
treated with VMAT were included. Their non-robustly
optimized (i.e. PTV-based) VMAT plans were copied and
their objectives were altered to create robustly optimized
(i.e. CTV-based) VMAT plans. Plans were re-optimized
with worst case robust objectives for the targets. Hence,
the actual given dose was estimated by calculating the
dose on 35 daily cone beam CT (CBCT) scans after position
correction, deforming the dose distributions to the
planning CT and summing doses from all fractions. All
estimated actual given dose distributions were inspected
and approved by a physician with experience in head and
neck radiotherapy. Treatment plan quality was evaluated
using dosimetric parameters and normal tissue
complication probability (NTCP) values using previously
published models for tube feeding dependence, grade 2-4
dysphagia, xerostomia and sticky saliva, 6 months after
treatment.
Results
Robustly optimised plans resulted in a lower actual given
dose estimation for all organs at risk (figure 1). The
difference in dose distribution and dose volume
histograms of a typical case are displayed in figure 2. The
average NTCP values were lower for the robustly
optimized plans by 0.5% (p = 0.07) for tube feeding
dependence, 1.8% (p = 0.04) for dysphagia, 3.0% (p = 0.01)
for xerostomia and 1.1% (p = 0.02) for sticky saliva.
Moreover, target dose conformity slightly improved using
robustly optimized VMAT optimization.
Conclusion
Robust CTV-based VMAT optimization in head and neck
patients resulted in improved estimated actual given dose
distributions with lower normal tissue dose and equal
target coverage compared to non-robustly optimized
plans. This is the first study to compare robustly optimized
photon plans to non-robustly optimized photon plans in
terms of dose accumulation using daily CBCT images. The
differences in dose are deemed clinically relevant and are
expected to lead to an improved method of patient
selection for proton therapy.