Session Item

In conventional EBRT a single treatment plan is generated pre-treatment, and delivered in each fraction. With this approach, dose spikes entering healthy tissues are in all fractions positioned in (approximately) the same locations. Therefore, planned doses in these spikes have to be kept low, possibly limiting reduction of doses in radiosensitive OARs. In this study, we propose and evaluate a novel SBRT approach, using automated planning to generate for each fraction a different treatment plan to reduce accumulated OAR doses (‘per-fraction planning’, PF planning). PF plans are all generated prior to treatment in a sequential procedure that always takes into account dose in already generated plans.

Planning CT-scans of ten prostate SBRT patients, previously treated with 4x9.5 Gy, were included. An in-house application for fully automated, non-coplanar treatment planning with integrated beam angle optimization (BAO) was used to compare PF planning with the conventional one-plan approach. For conventional planning, a 12-beam non-coplanar IMRT plan with individualized beam angles was generated for each patient. Different 12-beam configurations could be automatically selected for the 4 PF plans. PF plans were sequentially generated by adding dose to already generated PF plan(s), such that accumulated doses fulfilled all soft and hard planning constraints for PTV, OARs and other normal tissues. Moreover, each PF plan separately also obeyed all (hard constraints)/N, with N the number of fractions, while some violations of normal tissue soft constraints were allowed in individual PF plans.

In PF planning, the optimizer always selected different beam setups for the 4 fractions, resulting in different patterns of dose spikes ‘leaking’ from the PTV into healthy tissues, with low spike doses in accumulated plans (Fig.1, right panels), similar to conventional planning (Fig. 1, left panel). PTV doses in PF planning were acceptable and highly similar to those in conventional treatment, while fulfilling all OAR and PTV hard constraints. PF accumulated plans showed improved OAR sparing compared to conventional, especially for bladder and also for rectum (Fig. 1 and Fig.2); reductions in bladder Dmean of 24.5% (range: 13.5-35.3) and in rectum Dmean of 6.3% (range: -7.2-18.5). Patient whole body (normal tissue) dose was similar (Fig.2).

The proposed per-fraction planning with different, dosimetrically matched plans for the four SBRT fractions, resulted in reduced bladder and rectum doses compared to the conventional one-plan approach, for similar PTV and whole body patient doses. Extension of the methodology to on-line adaptive re-planning based on daily acquired CT- or MR-scans is feasible and a topic for further research. 

Figure 1: Dose distributions for one example patient for conventional planning and per-fraction planning (PF).

Figure 2: Population average DVHs of the total plans.