Evaluation of the effect of beam arrangements and establishment of treatment planning models in intensity modulated radiation therapy of head and neck cancers

Abstract

Introduction: Intensity modulated radiotherapy (IMRT) has been the main treatment modality for many head and neck (H&N) cancers. This study aims to provide guidance for planners in the treatment planning of head and neck cancers treated by IMRT. The guidance refers to the suggestion of optimal beam arrangement, the feasibility to escalate tumour dose and the determination of reference dose to the organs at risk (OARs). Methodology: Study 1: 5 types of H&N cancers were included and a total 119 patients previously treated with IMRT were recruited. 5 plans of different beam arrangement methods were optimized for each case, including equal spaced beam (ESB), coplanar beam angle optimization (BAOc), non-coplanar beam angle optimization (BAOnc), 2 arcs volumetric modulated arc therapy (VMAT2) and 3 arcs volumetric modulated arc therapy (VMAT3). Apart from the dose volume parameters, a "figure-of-merit" known as uncomplicated target conformity index (UTCI) was used to rank the beam arrangement method in each type of cancers. ANOVA with repeated measures was used to rank the plans according to UTCI. Study 2: 25 NPC cases (stage T3-4, N0-1) were recruited. With the same prescription of the planning target volumes (PTVs) and the planning goals of the OARs, 3 IMRT plans of different gross tumour volume (GTV) doses (76 Gy, 78 Gy and 80 Gy) were optimized using the BAOnc for each case. Paired sample T test was used to determine any statistical increase of OARs dose in the GTV dose between the dose escalated plans and the reference plan. Study 3: 70 cases of NPC patients (45 in training dataset, 25 in validation dataset) were retrieved from the database of a local hospital and a hypothetical IMRT plan was computed for each patient. Multiple regression analysis was carried out using the OAR dose parameters as the dependent variables and the anatomical parameters such as the distance and the overlapping volume between the OARs and PTVs as the independent variables. External validity of the multiple regression models was evaluated using the validation dataset. Results: Study 1: All treatment plans met the dose requirements for the PTVs and OARs. The OARs for the evaluation included brain stem, spinal cord, lens, optic nerve, optic chiasm, eyeball, pituitary, parotid, cochlea, temporal lobe and brachial plexus. The UTCI favoured the use of BAO and VMAT methods. BAOnc offered the best chance for OARs sparing. However, if treatment time was included into consideration, VMAT plans would be recommended for cancers of nasopharynx (VMAT3), oral cavity and larynx (VMAT2). Study 2: The dose to most OARs showed no statistical increase in the GTV dose escalated plans except lens, temporal lobe and pituitary. Despite the dose to temporal lobe and lens were increased, the dose did not exceed their tolerance even for the plans with GTV dose = 80 Gy. Only the pituitary gland demonstrated dose above its tolerance. Study 3: A total of 11 multiple regression equations, one for each OAR dose parameter, were formulated. The adjusted R2 value of the multiple regression models ranged from 0.916 for the brain stem to 0.436 for the lens. All multiple regression equations passed the validation to test the reliability of the standard error of the estimates except the eyeball and the lens. Conclusion: With regard to the optimal beam arrangement for H&N cancers treated by IMRT, VMAT was recommended for the cancer of nasopharynx, oral cavity and larynx; and BAOc for the cancer of maxillary sinus and parotid. In the dosimetric study for GTV dose escalation of NPC, most OARs tolerated the increased dose except the pituitary gland, which demonstrated dose beyond it tolerance in the GTV dose was escalated to 80 Gy. It is suggested that further study for more accurate case selection is needed, so that dose escalation will only be performed in those cases that weigh higher for the local control over the protection of pituitary gland. Lastly from the result of study 3, multiple regression models have been demonstrated to be able to determine reference OARs dose to guide the IMRT optimization.