Dose escalation for non-small cell lung cancer: analysis and modelling of published literature.

PURPOSE: To review the published clinical data on non-small cell lung cancer treated with radical radiotherapy to confirm a dose-response relationship as a basis for further dose-escalation trials. METHODS: Twenty-four published clinical trials were identified, 16 of which - with 29 different standard, hyper- and hypofractionated treatment schedules - were analysed. Prescription doses were converted to biologically-equivalent dose (BED), with a correction for repopulation. Disease-free survival data were corrected for the stage profile of each cohort to allow better comparison of results. We also analysed moderate (grade II and III) lung and oesophageal acute toxicity related to the corrected BED delivered to the tumour. RESULTS: The clinical data analysed showed good agreement between the observed and modelled disease-free survival at 2 years when compared to the published models of Fenwick (correlation coefficient 0.525, p=0.003) and Martel (correlation coefficient 0.492, p=0.007), indicating a clear tumour dose-response. In the normally fractionated treatments (∼ 2 Gy per fraction), improved disease-free survival was generally observed in the shorter schedules (maximum around 6 weeks). However, the best outcomes were obtained for the hypofractionated schedules. No systematic relationship was seen between prescribed dose and lung or oesophageal acute toxicity, possibly due to dose selection depending on V(20) or MLD in some studies and the diversity of the patients analysed. CONCLUSIONS: We have demonstrated a dose-response relationship for NSCLC based on clinical data. The clinical data provide a rational basis for selection of dose escalation schedules to be tested in future randomised trials