Session Item

For intensity modulated proton therapy (IMPT) planning, robust optimization (RO) and evaluation is used to compensate for worst case uncertainties in dose deposition due to translational changes in setup and stopping power (range). The two uncertainties can be either coupled or uncoupled in the optimization, but currently there are no clinical consensus regarding the optimal approach. Combining the uncertainties will potentially lead to unnecessary high doses to organs at risk (OAR). On the other hand, the RO does not take anatomical changes and e.g. flexion or rotation into account. 

In this study we evaluated coupled vs. uncoupled uncertainty scenarios in RO IMPT plans in head and neck cancer (HNC) patients with respect to normal tissue complication probability (NTCP) and robustness against re-planning.

21 consecutive oropharyngeal HNC patients treated with IMPT were selected. They were treated with 3 dose levels: 66-68/60/50Gy in 33-34fx using 4-7 fields and a 5cm range shifter. The clinical plans were optimized using robustness parameter combinations of setup uncertainty of ± 4mm in all cardinal directions and ± 3.5% range uncertainty - 14 coupled scenarios in total. New plans were re-optimized for similar target coverage with uncoupled robustness parameters - 8 uncoupled scenarios in total.

The difference in NTCP of ≥ Grade 2 dysphagia and ≥ Grade 2 xerostomia for the two RO approaches were recorded.

Fifteen of 21 patients had been re-planned during the treatment course (median fx11 (range: fx7-fx24)) due to insufficient target coverage. On the re-plan CT the target coverage from the two RO approaches was compared on size of largest cold spot volume and DVH parameters. Six of 21 patients were not re-planned and target coverage was compared on the last control CT in the final week of treatment.

Test for significant difference (p<0.05) were performed with Wilcoxon signed-rank test.

For initial planning the uncoupled plans showed a significant lower NTCP of median -0.4% (range: -1.2%;0.4%) for dysphagia and median -0.2% (range: -1.1%;0.2%) for xerostomia compared to the coupled scenarios.

Figure 1 shows the absolute size of the largest connected cold volume receiving less than 95% of the prescribed dose for the 3 CTV's when the nominal plans were re-calculated on re-plan CT or last control CT. The underdosage was significantly larger for the uncoupled RO.

Figure 2 depicts the relative volume receiving 95% of the prescribed dose from the re-calculation taking range uncertainty into account (worst case of 2 scenarios).

Initial planning with uncoupled RO yielded lower NTCP due to lower OAR dose. However, the uncoupled RO plans were less robust to anatomical changes requiring re-planning during the treatment course. Contemporary planning is not optimal as the probable scenarios of anatomical changes is not incorporated in the treatment planning systems, but using the improbable combination of uncertainty scenarios compensate somewhat for this.