On-treatment image-guided radiotherapy for central lung tumours

Central lung tumours are notoriously diï¬cult to treat with conventional or high dose per fraction radiotherapy regimens due to abutting and overlapping radiosensitive organs-at-risk. Using two patient cohorts, this thesis explores the scope of isotoxic dose escalation possible for central lung tumours due to new techniques now technically feasible on magnetic resonance (MR) linear accelerators (linacs), for locally-advanced (LA), and early-stage stereotactic radiotherapy, non-small cell lung cancer patients. Real-time MR imaging of orthogonal slice pairs or radiographically imaged ï¬ducial markers allow target tracking during radiotherapy treatment delivery. In-silico modelling of a tracking technique based on primary tumour centroids was performed, using varying clinical target volume (CTV) to planning target volume (PTV) margins based on ideal, accurate and less accurate tracking. The resulting accumulated dose distributions were assessed for both target coverage and dose to surrounding critical structures, to fully quantify the level of gains that real-time tracked lung radiotherapy on MR-linacs may provide. A clinically signiï¬cant increase in mean isotoxically prescribed tumour dose of 3.2 - 12.4 Gy was achieved for LA patients, resulting from the increased separation of target volumes and surrounding critical structures through the use of reduced target volume margins. Whilst reduced target margins resulted in larger reductions in minimum target volume doses compared to the standard internal target volume (ITV) approach when accumulated across the respiratory cycle, due to the large dose escalations achieved during isotoxic planning, an increase in absolute target doses still prevailed. Surrounding critical structures experienced small increases in maximum doses, causing planning tolerance violations. However these were of a similar magnitude (~0.5 Gy) and frequency as measured for the standard technique. Thus the tolerance violations were not clinically relevant, demonstrating the potential for safe implementation of the presented technique. Assessment of plans produced on images acquired 7 - 22 days prior to treatment delivery demonstrate notably reduced target coverage compared to plans created on the day of treatment, suggesting that daily plan adaption would result in optimal treatment plans. However, for 40% of patients, a higher target dose was achievable using âoldâ experimental plans compared to the standard ITV technique on up-to-date imaging, demonstrating a beneï¬t of the proposed technique for some patients. In-silico modelling predicts that tracked delivery of isotoxically dose escalated plans will result in a clinically meaningful increase in two-year disease-free survival for central locally-advanced NSCLC, while respecting current normal tissue dose-limits even allowing for movement eï¬ects.