A split-organ delineation approach for dose optimisation for intensity-modulated radiotherapy for advanced T-stage nasopharyngeal carcinoma

Aims: To assess the dosimetric effect of using a split-organ delineation approach during intensity-modulated radiotherapy (IMRT) treatment planning for advanced T-stage nasopharyngeal carcinoma (NPC). Materials and methods: Twenty NPC patients with T3-4 tumours were studied. A reference (REF) IMRT plan was generated based on a standard treatment planning protocol, with a set of user-defined dose constraints for optimisation. An investigative (INV) IMRT plan was then generated based on the same protocol, but treating several organs at risk (OARs; parotid glands, temporal lobes, cochlea, auditory nerves and planning organ at risk volume [PRV] of the brainstem) as split organs consisting of target-overlapping and non-target-overlapping sub-segments. These sub-segments were assigned independent dose constraints. The REF and INV plans were compared with respect to target coverage and OAR sparing. Target coverage was evaluated by the D-min (minimum dose), V-66/V-60 (percentage volume of gross target volume [GTV]/planning target volume [PTV] receiving 66 Gy/60 Gy), target conformity index (CI), and tumour control probability (TCP). The sparing of OARs was evaluated by the commonly used dose end points for the respective OAR, and normal tissue complication probability (NTCP). Results: For PTV coverage, the INV plan was superior to the REF plan in terms of Dmin (P = 0.000), CI (P = 0.005) and TCP (P = 0.002). This is attributed to an increase in dose to the PTV-OAR overlapping sub-segments. Regarding the sparing of OARs, there was a significant reduction in the mean dose of the parotid glands (P = 0.002), and a slight, but non-significant, increase in NTCP of the temporal lobes, cochlea and brainstem. Conclusions: Using a split-organ delineation approach in IMRT treatment planning for advanced T-stage NPC, a significant improvement in the target coverage and TCP could be achieved, whereas the mean dose of the parotid was reduced significantly. There was insignificant change in the NTCP of the temporal lobe, parotid gland, cochlea and brainstem, but a significant change in the NTCP of the auditory nerve. The approach provides the planner extra room to manipulate the dose constraints during optimisation, and to obtain the desired result in less attempts. This approach also has the potential to be used in a broader context for IMRT planning for other tumour sites.