Quantification of the uncertainties within the radiotherapy dosimetry chain and their impact on tumour control

BACKGROUND AND PURPOSE: Dose delivered during radiotherapy has uncertainty arising from a number of sources including machine calibration, treatment planning and delivery and can impact outcomes. Any systematic uncertainties will impact all patients and can continue for extended periods. The impact on tumour control probability (TCP) of the uncertainties within the radiotherapy calibration process has been assessed. MATERIALS AND METHODS: The linear-quadratic model was used to simulate the TCP from two prostate cancer and a head and neck (H&N) clinical trial. The uncertainty was separated into four components; 1) initial calibration, 2) systematic shift due to output drift, 3) drift during treatment and 4) daily fluctuations. Simulations were performed for each clinical case to model the variation in TCP present at the end of treatment arising from the different components. RESULTS: Overall uncertainty in delivered dose was +/−2.1% (95% confidence interval (CI)), consisting of uncertainty standard deviations of 0.7% in initial calibration, 0.8% due to subsequent calibration shift due to output drift, 0.1% due to drift during treatment, and 0.2% from daily variations. The overall uncertainty of TCP (95% CI) for a population of patients treated on different machines was +/−3%, +/−5%, and +/−3% for simulations based on the two prostate trials and H&N trial respectively. CONCLUSION: The greatest variation in delivered target volume dose arose from calibration shift due to output drift. Careful monitoring of beam output following initial calibration remains vital and may have a significant impact on clinical outcomes.