PTV dose prescription strategies for SBRT of metastatic liver tumours.

Purpose: Recently we have demonstrated that our in-house developed algorithm for automated plan generation for fully non-coplanar SBRT of liver patients (designated Cycle) yields plans that are superior to conventionally generated plans of experienced dosimetrists. Here we use Cycle in the comparison of plans with prescription isodoses of 65% or 80% of the isocentre dose.

Methods: Plans were generated using CT-data of 15 previously treated patients. For each patient, both for the 65%- and the 80% strategy, Cycle was used to generate a plan with the maximum isocentre dose, D(isoc), while strictly obeying a set of hard constraints for the organs at risk (OAR). Plans for the two strategies were compared using D(isoc), D(PTV,99%) (the minimum dose delivered to 99% of the PTV), and the generalised equivalent uniform dose, gEUD(PTV)(a), for several values of the parameter a. Moreover, for the OARs, the distance to the constraint values was analysed.

Results: The 65% strategy resulted in treatment plans with a higher D(isoc) (average 17.6%, range 7.6-31.1%) than the 80% strategy, at the cost of a somewhat lower D(PTV,99%) (average -2.0%, range -9.6% to 9.3%). On average, voxels with a dose in the 65% strategy, lower than the minimum PTV dose in the 80% strategy, were within 0.2cm from the PTV surface. For a-10, the 65% strategy yielded on average a significantly (P<0.01) higher gEUD(PTV)(a) than the 80% strategy, whereas for highly negative a-values the 80% approach was slightly better, although not significantly. Large variations between patients were observed. Generally, for the OAR the approach to the constraint levels was similar for the two strategies.

Conclusion: On average, PTV dose delivery is superior with the 65% strategy. However, apart from the isocentre dose, for each applied PTV dose parameter at least one patient would have been better off with the 80% dose prescription strategy.