Purpose: To test the feasibility of a new inverse planning technology based on the Hybrid Inverse treatment Planning and Optimisation (HIPO) algorithm for image-guided cervical cancer brachytherapy in comparison to conventional manual optimisation as applied in recent clinical practice based on long-term intracavitary cervical cancer brachytherapy experience.
Materials and methods: The clinically applied treatment plans of 10 tandem/ring (T/R) and 10 cases with additional needles (T/R+N) planned with PLATO v14.3 were included. Standard loading patterns were manually optimised to reach an optimal coverage with 7 Gy per fraction to the High Risk CTV and to fulfil dose constraints for organs at risk. For each of these patients an inverse plan was retrospectively created with Oncentra GYN v0.9.14. Anatomy based automatic source activation was based on the topography of target and organs. The HIPO algorithm included individual gradient and modification restrictions for the T/R and needle dwell times to preserve the spatial high-dose distribution as known from the long-term clinical experience in the standard cervical cancer brachytherapy and with manual planning.
Results: HIPO could achieve a better target coverage (V100) for all T/R and 7 T/R+N patients. Changes in the shape of the overdose volume (V200/400) were limited. The D(2 cc) per fraction for bladder, rectum and sigmoid colon was on average lower by 0.2 Gy, 0.4 Gy, 0.2 Gy, respectively, for T/R patients and 0.6 Gy, 0.3 Gy, 0.3 Gy for T/R+N patients (a decrease from 4.5 to 4 Gy per fraction means a total dose reduction of 5 Gy EQD2 for a 4-fraction schedule). In general the dwell times in the additional needles were lower compared to manual planning. The sparing factors were always better for HIPO plans. Additionally, in 7 T/R and 7 T/R+N patients all three D(0.1 cc), D(1 cc) and D(2 cc) for vagina wall were lower and a smaller area of vagina was covered by the reference dose in HIPO plans. Overall loading times in the tandem, the ring and the needles, as well as dose distribution, were largely preserved with adaptations performed due to specific topographical variations, in particular in lateral and caudal directions.
Conclusions: Inverse planning based on the HIPO algorithm can produce treatment plans for cervical cancer brachytherapy which are comparable to plans based on manual optimisation as applied in clinical practice. It is essential to take into account the spatial dose distribution in addition to the DVH-based constraints. The proposed inverse planning concept is feasible for improving the therapeutic ratio and limiting substantial high-dose regions around needles.