Biologically effective doses in medium dose rate brachytherapy of cancer of the cervix.

The amount of dose reduction on changing from low dose rate (LDR) brachytherapy to medium dose rate (MDR) or high dose rate (HDR) afterloading has been the subject of much debate. The magnitude of reduction depends, together with other possible factors, on two radiobiological parameters: the alpha/beta ratio and the half-time of repair of the relevant tissues. In an attempt to extract these radiobiological parameters for the late rectal complications observed in our previously published clinical results four different schedules using MDR and one using LDR are analyzed. The percentage incidence of complications was a function of increasing biologically effective dose (BED), but would yield nonsense scattergrams if plotted against raw total dose. In addition, for three other published MDR series, three LDR series, and two HDR series, the incidence of rectal complications is plotted against BED to examine the predictive potential of using BED as the surrogate of total dose. Our own results were published in 1996, consisting of 102 patients treated at the LDR of 0.44 Gy/hr and 88 patients treated by four different schedules using an MDR of 1.6-1.7 Gy/hr. Follow-up is at least 3 years in all schedules. The linear quadratic formula including the "g" dose rate factor was used to analyze them, assuming exponential repair of the repairable beta term. First, multivariate and profile likelihood analyses were carried out to obtain estimates of alpha/beta and T1/2 for rectal late responding tissues. Then graphs of incidence of rectal complications vs. BED were constructed, assuming alpha/beta = 3 Gy and T1/2 = 1.5 hr, values which had not been contradicted by the multivariate analysis. Graphs were drawn both for "all grades including mild reactions" (grades 1 + 2 + 3) and for "serious" complications (grade 3 in our system). In addition, other published cervical brachytherapy series were reviewed, with calculation of their BEDs if not published by the authors. It was necessary to review and compare their grading systems, so that "mild and moderate" (grades 1 and 2) could be contrasted with "serious" (grades 3 and 4 or 5 in various systems). Comparisons were made with other published results, including three LDR, three MDR, and two HDR series spanning from 1982 to 1997. The BEDs at which the incidence of rectal complications rose above the arbitrary level of 10% were compared for all three ranges of dose rate. The multivariate analysis gave estimates of alpha/beta and T1/2 which were not significantly different from 3 Gy and 1.5 hr, respectively, so these values were used to compute the BEDs for the subsequent comparisons. It was found that the graphs of incidence of rectal complications for "all grades including mild" agreed rather better between all series than might have been expected, within a provisional (10%) threshold BED of range 100-123 Gy3 (60-74 Gy given as 2 Gy fractionated external beam or as LDR). The dose-response curves diverged above these values, as expected until common grading systems such as SOMA/ LENT become more widely used. For "serious" complications the 10% incidence occurred at a median BED of 140 Gy3 (84 Gy given as 2 Gy fractionated external beam or as LDR), range 124-155 Gy3. The use of BED (or extrapolated response dose), assuming alpha/beta = 3 Gy and T1/2 = 1.5 hr, instead of total dose, enabled incidence of late rectal complications in cervical brachytherapy with LDR, MDR, and HDR to be plotted in a reasonably consistent way. This does not mean that those parameter values have been definitively determined, but they appear to be provisional values that may be of use in comparing the expected effects of new schedules until better values are obtained from greater use of common grading systems.