Compensation for changes in dose-rate in radical low-dose-rate brachytherapy: a radiobiological analysis of a randomised clinical trial

Abstract

Background and purpose This study reanalysed the results of the Cs-137 low-dose-rate brachytherapy trials for stage I and II cervix carcinoma at the Christie Hospital, Manchester, UK, in order to quantify the clinical outcome as a function of dose, and to extract radiobiological parameter values by modelling the data for local control and morbidity. Patients and methods Kaplan–Meier survival curves and Cox regression analyses were used to analyse the time to event data. Linear-quadratic (LQ) analysis was also used in a mixture model, incorporating a half-time for repair, a time factor, and a heterogeneity function between patients. Full 5-year follow-up data were available for 339 patients receiving Cs-137 doses between 60 and 75 Gy delivered at 1.4–1.8 Gy/h, and 178 patients receiving a Ra-226 dose of 75 Gy at 0.5 Gy/h, using two insertions 7–10 days apart. Results With the increased dose-rate, a dose reduction between 20 and 25% was required to achieve a similar morbidity rate. This reduction had a detrimental effect on tumour control, by about 15% points. Unexpectedly, this loss in local control did not lead to a decrease in cancer-specific survival. For both tumour control and complications a high α/ β and short half-time for repair best fitted the data, suggesting that consequential late reactions may be responsible for much of the bowel and urinary morbidity after these short treatments. The variability in response between patients was greater (CV 40%) for morbidity than for tumour control (CV 17%), probably reflecting the greater variation in dose at the target tissue. There was no significant dependence on overall treatment time detected over the 7–10-day range of these treatments. Conclusions The therapeutic ratio was somewhat less for the higher dose-rate, in agreement with radiobiological expectations, although cancer-specific survival was inexplicably unchanged. The LQ-parameter analysis suggests that high α/ β ratios and/or short repair half-times are applicable for both tumour and normal tissue responses in these treatments.